Post from Andy Nelson (a.nelson@cgiar.org):
I’ve made a start on a dataset that relates to the poverty / agriculture question and which should help us to get a little closer to a commonly agreed methodology for linking poverty numbers to agricultural production/consumption for the MP proposals and many other applications. I’d like to get things moving here and am happy to receive offers of collaboration/data/advice/condolences.
Following on from previous work by Stan et al in the PAGE report and David Raitzer in SE Asia, I have made an updated Value of Production grid at 10km resolution which represents the summed value (in USD) of production for around 120 crops per hectare.
- I used the 10km resolution Monfreda crop yield and area data for 175 crops (no livestock or timber products) to compute production in tonnes per 10 x 10 km pixel for the year 2000. http://www.geog.mcgill.ca/landuse/pub/Data/175crops2000/
- I then took the FAO national level production data for 2005 – matched them to the Monfreda layers as closely as I could and thus corrected the Monfreda data to get a 10km resolution production map for 2005 for 159 crops. There were 16 crops where I could not find a match between FAO and Monfreda. http://faostat.fao.org/site/567/DesktopDefault.aspx?PageID=567
- Then I took the FAO internationally comparable prices for the top 20 agricultural commodities per country. This includes livestock. These prices are described as follows “International commodity prices are used, to calculate the total value of each commodity produced by each country and subsequently used in the ranking of commodities and countries. They are applied in order to avoid the use of exchange rates for obtaining continental and world aggregates, and also to improve and facilitate international comparative analysis of productivity at the national level.”
The cross tabulation of prices against countries results in lots of gaps since we only have 20 prices per country. To resolve this, I summarized the data into 22 UN regions and computed the median price for each region based on the available country prices. If there were no countries in that region with price data for a given commodity I further aggregated to 5 UN super regions (continents) and computed median prices again. If there were no countries in these super regions with prices, I simply computed a global median price. These regional, super regional and world median prices were used to fill the gaps in the commodity / country cross tabulation, using the finest level of available price data. http://faostat.fao.org/site/339/default.aspx
The prices are very constant across countries in a region so I feel that this is a reasonable method. However, I would much prefer real price data for all crops. I used median prices to limit the effect of outliers on the regional price (eg. Camel meat is 1400USD per tonne everywhere except Kuwat where it is over 5000USD per tonne). - These 2005 international prices were multiplied by the 2005 production to give USD value of production per crop per 10km x 10km pixel. This was done for 122 crops. Not all crops appear in the FAO price data. Note that we have price data for most livestock products (23 in total) but do not have raster maps of livestock production to multiply them with. Does ILRI have such data that we could plug into this?
- We now have 10km x 10km maps of yield, area, production and USD value for 122 crops globally. These can be summed to give a total value of crop production (VofP) per pixel, even though we miss the livestock and timber products. They can also be used to compute the % contribution of each crop to the total value.
See maps below for total crop VofP per hectare (2005) and % of VofP from rice ? Again, there are no livestock or timber products here, just crops.
Possible next steps
- Locate better price data to avoid filling in gaps for crops that fall outside the top 20 in each country. Does anyone know if FAO has more data than they present on the FAOSTAT website?
- Locate spatial livestock production data and add to the analysis. FAO has national livestock production data. If there are no global production maps, can we instead distribute the FAO production based on livestock farming systems?
- Locate spatial timber production data and tabular timber value data(?)
- Any other commodities to add?
I think that this method is straightforward and plausible. We only need to improve the input data or fill in gaps to make a more comprehensive map of V of P. The following points are more problematic.
- Link this data to poverty. How to assign a share of the production value to the rural poor in each pixel? The crudest method is to multiply the total value by the poverty rate and then divide by number of poor to give USD of production per year per person in poverty per pixel. It would be far better to break out the value based on knowledge of farm size distribution / crops typically cultivated by the poor etc. etc. in each country.
- The output of this should get us much closer to an idea of the contribution or importance of each crop to the poor in each country/farming system/aez.
- What about consumption data? This current method only accounts for production and the rural poor. We may need a totally different approach for consumption and the urban poor. Any thoughts?
- What about costs? We only look at the production value. We should try to subtract the land rent, input costs, labour costs etc. Does anyone have any data that we can bring to bear on this?
- What about fat/protein/calories per commodity per capita? Can this be usefully mapped? This seems straightforward. Do we have the data ?
And, here are two attachments:
- analysis_steps.xls (27 KB)
Details on the input data and where I think improvements can be made - poverty_and_crops.xls (956 KB)
The results so far in terms of value of production by the poor (production per pixel x poverty rate per pixel) and poverty weighted value of production by the poor (production per pixel x % of the global poor per pixel). The results are summarized as graphs per continent showing the crop value, ranked from most valuable to least valuable.
Look forward to your comments.
Andy Nelson
IRRI
