Food and nutrition security: Early warning systems for commodity markets
The state of food security and the stability of the food system can be summarized and measured by food price pressure as well as food price volatility. In this way changes in food prices and their volatility can be used as early warning indicators. SUSFANS researchers of the International Institute for Applied Systems Analysis (IIASA) illustrate in a new report the capacity of an econometric system to provide such services.
The price and price volatility forecasting model constitutes SUSFANS’s operational early warning system for agricultural commodity markets, and contributes to the wider SUSFANS toolbox. The AgriPrice4Cast model is providing seasonal prices based on short-term yield forecasts for agricultural crops. The seasonal price forecasts allow for the planning of emergency measures in cases of harvest outages, financial market or macro-economic shocks in the world and/or designing storage and other stabilization measures.
An early warning system for the most relevant agricultural commodities
In a new SUSFANS deliverable D8.4, they demonstrate an early warning system (EWS) for the most relevant agricultural commodities. The design of the early warning system in the SUSFANS project is complementary to the ones already existing.
In particular, they would like to mention the initiatives of the G20 Action Plan on Food Price Volatility and Agriculture in dedicated forums: Agricultural Market Information System (AMIS) and the Rapid Response Forum, GEO Global Agricultural Monitoring Initiative (GEOGLAM) for market and production international monitoring, and risk management tools, such as the Platform for Agricultural Risk Management (PARM), and the initiatives including the Global Agriculture and Food Security Programme (GAFSP).
The SUSFANS forecasting system is complementary to the early warning systems mentioned above, which are mostly concentrated around production and yield forecasting, as it includes a large set of financial and macro-economic forecasting variables.
Broader Analysis with more mainstream commodities
In their previous paper, SUSFANS deliverable 8.3, the SUSFANS research team presented a comprehensive modelling framework aimed at obtaining short-term forecasts. After first testing with their case study on commodity coffee, they expanded the analysis to more mainstream commodities, which are also more relevant for the EU food market. Short-term forecasts had been performed one to twelve months ahead of commodity prices and predictions were generated in addition to Arabica coffee for the following three commodities: wheat, soybeans and corn.
These commodities were selected as showcases, because wheat is the "bread" crop of Europe occupying almost half of total cropland in use and producing 160 million tons per year, which corresponds to 50 percent of total coarse grain output.
Soy was used as an iconic import crop for Europe with half of total oilseed meal consumption made up of soymeal imports.
Corn is another iconic European crop covering 20 percent of total coarse grain output.
In their analysis in Deliverable 8.4, the SUSFANS researchers maintain a large number of univariate and multivariate time series models, including specifications that exploit information about market fundamentals, macroeconomic and financial developments and climatic variables.
Address model uncertainty
A comprehensive set of forecast averaging tools was implemented to explicitly address model uncertainty. Results indicate that variables measuring market fundamentals and macroeconomic developments (and to a lesser extent, financial developments) contain systematic predictive information for out-of-sample forecasting of commodity prices. The market fundamentals in terms of physical production played an increasingly important role compared to the original case study of Arabica coffee.
Within the SUSFANS framework the AgriPrice4Cast model can serve multiple purposes. One of the original ideas was to develop a price and price volatility estimator to inform long-run food security models and provide early warning.
This is motivated by the fact that it is less the slowly moving price trends which create episodes of food insecurity, but it is more related to extreme events including episodes of price volatility and price uncertainty both as an endogenous factor due to the shock as wells as the main external shock on the respective food system.
Together with the market pressure index from either the GLOBIOM or also MAGNET model, episodes and subsequent states of food insecurity in vulnerable geographies can, thus, be better characterized.
Also for long-run assessment of the stability of FNS a price volatility reaction function can serve valuable insights provided that the overall market pressure conditions are replicable in such a longer-run future.
It is also recommended that the current statistical methods determining the number of hungry and food insecure people should be enriched by indicators of price volatility. Targeted research in this direction should be conducted with partners such as the FAO or IFAD to generate evidence from analysis of the impact of volatility on household or individual level and subsequent up-scaling methodologies. Following this line of research the collection of high frequency price data from a multitude of market locations would be a prerequisite to roll out such a new market and food security system. The usefulness of crowd-sourcing tools to collect highly geographically and temporally resolved input data for the EWS cannot be over-emphasized. Finally, the EWS function of the AgriPrice4Cast tool could also be used by the humanitarian community to optimize their operations.