Risk of Hunger Under Climate Change, Social Disparity, and Agroproductivity Scenarios

Abstract

Considering the projected population growth in the twenty-first century, some studies have indicated that global warming may have negative impacts on the risk of hunger. These conclusions were derived based on assumptions related to social and technological scenarios that involve substantial and influential uncertainties. In this paper, focusing on agrotechnology and food access disparity, we analyzed food availability and risk of hunger under the combined scenarios of food demands and agroproductivity with and without climate change by 2100 for the B2 scenario in the Special Report on Emissions Scenarios. The results of this study suggest that (1) future food demand can be satisfied globally under all assumed combined scenarios, and (2) a reduction of food access disparity and increased progress in productivity are just as important as climate change mitigation for reducing the risk of hunger.

Appendices

Appendix I: IPCC SRES Scenarios

There are four basic scenarios in the SRES report: A1, A2, B1, and B2. With regard to the population prospects, there are only three different trajectories because scenarios A1 and B1 have the same population trajectory. These two scenarios assume a higher economic growth around the world and lower birth rates in developing countries. As a result, they have the lowest population trajectory of the scenarios. The A2 scenario assumes self-reliance regions, lower trade flows, and uneven economic growth. Reflecting international disparities, a higher birth rate in developing regions and the largest population among the scenarios is assumed. The B2 scenario describes intermediate economic growth and moderate population growth, and the fertility rates were assumed to converge to the same level as the UN 1998 medium scenario.

Appendix II: Feed Demand for Animal Production

In this study, we assumed four types of animal products: bovine meat, pig meat, poultry, and milk. Alexandratos [38] provides a simple equation to estimate the feed demand for animal production. However, the intensity of feeding varies significantly by region: in some regions, the products come from non-grain-fed animals, whereas other regions feed high-calorie cereals intensively. We assume that two feed intensity vectors, V _e and V _i , are the required feed for unit production of the various types of animal products in extensive and intensive feeding systems. Assuming a region-specific parameter α, the regional feed demand Q is calculated using the following equation:

$$ {Q_f}=\left( {\left( {1-\alpha } \right)\cdot {V_e}+\alpha \cdot {V_i}} \right)\cdot X $$

(7)

where X is the production volume vector of animal products. V _e and V _i are assumed on the basis of Alexandratos’ study and reports from agricultural experiments in Japan. Using the data of Q _f and X in 2000 from FAOSTAT, α is estimated for each region.

We also assume that the feed consists of the eight kinds of crops considered in the land use analysis in this study. The regional share of primary crops for feed is fixed to the value in 2000.

Appendix III: Food Trade Model

In this model, the regional food demand is satisfied by domestic and imported products. The share of domestic product S _D is calculated by a binary logit model using the domestic price P _D and import price P _I .

$$ {S_D}=\frac{{\exp \left( {-{\theta_{d1 }}\cdot {P_D}+{\theta_{d0 }}} \right)}}{{\exp \left( {-{\theta_{d1 }}\cdot {P_D}+{\theta_{d0 }}} \right)+\exp \left( {-{\theta_{d1 }}\cdot \left( {1+\tau } \right)\cdot {P_I}} \right)}} $$

(8)

where τ is the tariff rate, θ _d1 is the scale parameter of the error term, and θ _d0 is the dummy parameter for domestic product. This equation means that as the domestic price decreases or the import price increases, the domestic share becomes high. The model parameters are estimated for each food item and region. Using the share and demand for all regions, the global demand for imports is calculated. The export share of region j in the global market, which is denoted as S _Ej , is estimated by a multinomial logit model using the domestic prices of export regions P _Dj .

$$ {S_{Ej }}=\frac{{\exp \left( {{\theta_{e1 }}\cdot {P_{Dj }}+{\theta_{ej }}} \right)}}{{\sum\nolimits_{{j\prime }} {\exp \left( {{\theta_{e1 }}\cdot {P_{Dj' }}+{\theta_{{ej\prime }}}} \right)} }} $$

(9)

where θ _e1 is the scale parameter of the error term and θ _ej is the dummy parameter for region j. The model parameters are estimated by the following process. First, scale parameters are estimated by solving the following equation for θ ₁:

$$ \sigma =-\frac{{{p_1}/{p_2}}}{{{Q_1}/{Q_2}}}\cdot \frac{{\mathrm{d}\left( {{Q_1}/{Q_2}} \right)}}{{\mathrm{d}\left( {{p_1}/{p_2}} \right)}}=\frac{{{p_1}\cdot {p_2}}}{{{p_1}+{p_2}}}\cdot {\theta_1} $$

(10)

where σ is the elasticity of substitution given by GTAP [27], p ₁ and p ₂ are the prices of two competitive items, and Q ₁ and Q ₂ are the demand volume of the items. For the domestic share model, θ _d1 can be estimated using domestic and import prices. For the export share model, we calculate θ ₁ for all price combinations of two export regions and θ _e1 is determined as an average of them. Second, the dummy parameters θ _d0 and θ _ej are estimated to represent the share in 2000 under the estimated scale parameters.

In this model, food trade depends only on the price and tariff rate. However, in this study, we fixed the prices of agricultural products at the 2000 level during the analysis period. Therefore, the share of domestic products and the export share in the global food market are also fixed to the share in 2000 if land shortage does not occur in the allocation process. If land shortage occurs, the shortage volume of production is reallocated through the export share model where the export regions are limited to regions with marginal land for cultivation.