The use of inorganic fertilizer and other purchased inputs remains very low amongst African smallholder farmers. A central hypothesis about why this is the case, is that fertilizer use is simply not profitable, given the effective price of fertilizer at the farm gate, as well as the farm gate price of farm outputs (e.g. maize grain).
In a very stylized way, input-output price ratios are known to increase as one moves further away from markets (Figure below). Input prices increase with distance as they incorporate the price of movement from the port or blending facilities to farm locations. Output prices, in contrast, decrease as one moves away from market centers, reflecting the costs of transport and intermediation by traders. Even if fertilizer use is profitable in areas with relatively good access to markets, as one moves further from market centers this profitability can be expected to decrease and beyond some threshold of remoteness it no longer makes economic sense to participate in input and output markets.
Few development practitioners would argue with the stylized story told above, and there is a wealth of empirical support for the general idea that remote areas have lower rates of market participation. The trouble is defining exactly how prices are changing over space. Where exactly are the critical remoteness thresholds beyond which market engagement is a non-starter? In order to answer this, we need to know something about the distribution of local prices for inputs and outputs.
The problem is that good quality, spatially explicit data on input and output prices are difficult to get, particularly in the data sparse environments of rural Africa. Part of what we are working on in TAMASA is coming up with ways to model the spatial patterns of prices across rural landscapes, given the data that do exist. A useful place to start a discussion of this activity is to review the current available of price data.
Current sources of available price data are listed in the table below. In terms of output prices, our emphasis has been on maize grain, but many other wholesale prices are also available. For input prices, we have focused on local prices of inorganic fertilizer. Our country focus has been on the TAMASA countries, i.e. Ethiopia, Nigeria and Tanzania.
These data often give a coherent picture about how prices in different regional markets are varying across time. For example, using weekly time-series data from RATIN, we observe how regional market prices vary across the growing season (figure below).
Spatio-Temporal Variation in Maize Grain Prices in Tanzania
Source: compiled from data from the Regional Agricultural Trade Intelligence Network (RATIN) for 2016
But what about the variation of prices over space at location which are not major urban centers? With enough market locations in spatially referenced datasets, in principle, we should be able to see how much of the observed spatial variation is explainable, using spatial covariates of theoretical importance and our assumptions about how markets are spatially integrated. The map below shows a simple prediction of wholesale maize grain prices for 2013, using data from the LSMS-ISA for Tanzania, and a simple spatial regression model that incorporates travel time to different market locations. Using just two covariates – travel time to Dar es Salaam and population density — we are able to explain about 40% of the variation in reported local maize prices.
Predicted Maize Prices in Tanzania
Unfortunately – and somewhat surprisingly – we do much worse when it comes to predicting fertilizer prices. One challenge we have encountered has been to try to distinguish the signal from the noise in price data. As an example of this, consider data on we collected recently (October 2017) via an SMS poll conducted by GeoPoll (figure below). When we examine the prices reported for the town of Njombe, we see enormous variation in the reported prices (and in availability) from different respondents. This suggests that either (a) reported data are very noisy, (b) there actually is tremendous price variation within local markets, or (c) both.
Wholesale maize and inorganic fertilizer prices reported for Njombe market, Tanzania, October 2017
Note: GeoPoll data, collected via SMS in early Oct-2017. Maize price: TSh / 100kg bag; Fertilizer prices: TSh / 25kg bag. NA = not available in this market. MM = Minjingu Mazao; YC = YaraMila Cereal; YW = YaraMila Winner.
Map showing locations of GeoPoll price data collected in Tanzania in October 2017:
We are currently experimenting with alternative data sources and prediction methodologies. Stay tuned for further updates!
Spatially distributed input and output price data for Ethiopia, Nigeria and Tanzania
|FEWSNet commodity prices||wholesale and retail prices on commodities, multiple markets per country||monthly, 1995-2017||Nigeria
(and other non-TAMASA countries)
|16 markets||output grain||website|
|GIEWS commodity prices||wholesale and retail prices on commodities, multiple markets per country||monthly, 1995-2017||Various||Depends on country||output grain||website|
|RATIN||wholesale prices for maize and other grains||weekly, 2016||Tanzania
(and other non-TAMASA countries)
|Arusha, Mbeya, Dar||output grain||website & by subscription|
|WFP||wholesale prices (maize, etc.)||2016||Tanzania||19 markets||output grain||website|
|AfricaFertilizer||Fertilizer wholesale prices; coverage is spotty; major quality issues||monthly but with lots of gaps||Tanzania, Nigeria||varies by year, product, country (many missing values)||inorganic fertilizer: many types||website|
|LSMS-ISA Tanzania||farmer reported prices; coverage is spotty; major quality issues||2008-09
|Tanzania||many||maize, inorganic fertilizer||requires data mining; raw data available|
|LSMS-ISA Nigeria||farmer reported prices; coverage is spotty; major quality issues||2010-11
|Nigeria||many||maize, inorganic fertilizer||requires data mining; raw data available|
|LSMS-ISA Ethiopia||farmer reported prices; coverage is spotty; major quality issues||2011-12
|Ethiopia||many||maize, inorganic fertilizer||requires data mining; raw data available|
|APS Tanzania||community prices at harvest time||2016, 2017||Tanzania||25 districts||maize, fertilizer||coming soon!|
|APS Nigeria||community prices at harvest time||2016, 2017||Nigeria||3 states||maize, fertilizer||coming soon!|
|APS Ethiopia||community prices at harvest time||2016, 2017||Ethiopia||varies by year||maize, fertilizer||coming soon!|
Jordan Chamberlin Camila Bonilla Cedrez