Extreme weather, food security and the capacity to adapt – the case of crops in China

Abstract

Three extreme weather scenarios are examined for agriculture in China in this study. One scenario assumes a year when every province has precipitation corresponding to the lowest level experienced in the province over the last three decades. Another scenario assumes the highest experienced precipitation for every province is happening; and the last one assumes that the most harmful level of precipitation on crops occurs for every province – whether too little or too much. We studied the role of autonomous adaptation by farmers and through markets as embodied in a computable general equilibrium model. The results show that observed extreme impacts of precipitation on crop harvests are not serious for China at national level. The maize harvest is the most negatively affected with a reduction of 4 % without adaptation and less than 1 % reduction with adaptation. However, the impacts within a province may be serious and even become worse with adaptation. Good harvests might not make farmers better off due to lower crop prices even though consumers benefit. Sensitivity analysis shows that the ability to adapt assumed in the analysis may not be present in the short term.

Appendix. The GRACE model with crop farming by province

In GRACE, the endowment of production factors, i.e. labour, capital and natural resources within each region and time period are exogenous. Labour can move freely among production activities within a region, whereas natural resources are activity-specific and cannot be reallocated among sectors. The newly formed capital and depreciation of capital in the previous period can be freely allocated among activities and the other part of capital is activity- specific. The model assumes full utilization of all available resources within China (and the rest of the world). Producers pursue profit maximization and consumers pursue utility maximization.

Trade is modelled as bilateral with substitution among regional contributions. The substitution elasticities are based on those in the MIT EPPA model (Paltsev et al. 2005). Income of a region includes the remuneration to primary factors of production (labour, capital and natural resources) and direct and indirect taxes.

Saving is a fixed share of total income by region. All savings are used to invest in the world economy such the expected rates of returns to capital change at the same rate for all regions. Within each region, the investment is allocated to production activities such that the rates of returns to the new capital are equalized while the capital stock already existing at the beginning of each period is assumed to be activity-specific even though 4 % of it is depreciated and becomes one part of regional savings. The rates of returns to capital are equalized among regions in the long run.

In the standard GRACE model (Aaheim and Rive 2005), there is only one representative producer to produce one type of product in a sector. In order to utilize GRACE to analyse the impacts of precipitation at the provincial level, the output of each of the three cereal crops is disaggregated into 31 provinces proportional to the crop harvests shares by province. By assuming production technology of each crop is the same as in the agricultural sector in provincial input–output tables⁶ (NBSC 2011), we then utilize the direct input shares in agricultural output (direct input coefficients) to derive preliminary input values of commodities and productive resources in each province. Then a weighted least squares method is adopted to determine provincial inputs such that the final inputs of each commodity or resource in a crop production sum up to the corresponding national level in GTAP v8 database. The method aims to minimize the objective function defined as the sum of squared differences between final and original direct input coefficients (input share in total output of a crop) weighted by provincial shares of crop harvests (NBSC 2008). In doing so, the input structure in a province with high share of crop output is much closer to the original one stated in the provincial input–output table than a province with low share of crop output.