Risks of Correlated Wheat Yield Losses in the Global Breadbaskets

As the 2008 food price crisis or the 2010 drought in Russia and other cereal producing areas have shown, it is mainly the confluence of crop losses in different parts of the world that cause immense price shocks and risks to global food security which can lead to famine and political instability. However, to date, little is known about the yield loss dependence structures of the global “breadbaskets”. In order to model the interdependencies between historically observed wheat yield deviations in 5 “breadbaskets” - USA, Argentina, India, China and Australia – the copula methodology is used, a statistical tool to build multivariate distributions by joining univariate marginal distributions. This study uses and compares three different ways of constructing multivariate copulas: vine copulas, ordered coupling using the mini-max approach and hierarchical structuring. Investigating the correlation structures of wheat yield deviations, both within and between “breadbaskets”, high correlations within and not very significant correlations between the regions were found. Comparing wheat production losses in the "breadbaskets”, it is shown that ignoring correlation structures within breadbaskets leads to a significant underestimation of risks of production losses. This is especially interesting for crop insurance schemes or crop price forecasting. On a global scale, it is shown that inter-regional risk pooling between the breadbaskets could improve crop insurance schemes and decrease post-disaster liabilities of governments and international donors.