Why do global long‐term scenarios for agriculture differ? An overview of the AgMIP Global Economic Model Intercomparison

Recent studies assessing plausible futures for agricultural markets and global food security have had contradictory outcomes. To advance our understanding of the sources of the differences, 10 global economic models that produce long‐term scenarios were asked to compare a reference scenario with alternate socioeconomic, climate change, and bioenergy scenarios using a common set of key drivers. Several key conclusions emerge from this exercise: First, for a comparison of scenario results to be meaningful, a careful analysis of the interpretation of the relevant model variables is essential. For instance, the use of “real world commodity prices” differs widely across models, and comparing the prices without accounting for their different meanings can lead to misleading results. Second, results suggest that, once some key assumptions are harmonized, the variability in general trends across models declines but remains important. For example, given the common assumptions of the reference scenario, models show average annual rates of changes of real global producer prices for agricultural products on average ranging between ?0.4% and +0.7% between the 2005 base year and 2050. This compares to an average decline of real agricultural prices of 4% p.a. between the 1960s and the 2000s. Several other common trends are shown, for example, relating to key global growth areas for agricultural production and consumption. Third, differences in basic model parameters such as income and price elasticities, sometimes hidden in the way market behavior is modeled, result in significant differences in the details. Fourth, the analysis shows that agro‐economic modelers aiming to inform the agricultural and development policy debate require better data and analysis on both economic behavior and biophysical drivers. More interdisciplinary modeling efforts are required to cross‐fertilize analyses at different scales.     

Projecting future crop productivity for global economic modeling

Assessments of climate change impacts on agricultural markets and land‐use patterns rely on quantification of climate change impacts on the spatial patterns of land productivity. We supply a set of climate impact scenarios on agricultural land productivity derived from two climate models and two biophysical crop growth models to account for some of the uncertainty inherent in climate and impact models. Aggregation in space and time leads to information losses that can determine climate change impacts on agricultural markets and land‐use patterns because often aggregation is across steep gradients from low to high impacts or from increases to decreases. The four climate change impact scenarios supplied here were designed to represent the most significant impacts (high emission scenario only, assumed ineffectiveness of carbon dioxide fertilization on agricultural yields, no adjustments in management) but are consistent with the assumption that changes in agricultural practices are covered in the economic models. Globally, production of individual crops decrease by 10–38% under these climate change scenarios, with large uncertainties in spatial patterns that are determined by both the uncertainty in climate projections and the choice of impact model. This uncertainty in climate impact on crop productivity needs to be considered by economic assessments of climate change.     

The future of food demand: understanding differences in global economic models

Understanding the capacity of agricultural systems to feed the world population under climate change requires projecting future food demand. This article reviews demand modeling approaches from 10 global economic models participating in the Agricultural Model Intercomparison and Improvement Project (AgMIP). We compare food demand projections in 2050 for various regions and agricultural products under harmonized scenarios of socioeconomic development, climate change, and bioenergy expansion. In the reference scenario (SSP2), food demand increases by 59–98% between 2005 and 2050, slightly higher than the most recent FAO projection of 54% from 2005/2007. The range of results is large, in particular for animal calories (between 61% and 144%), caused by differences in demand systems specifications, and in income and price elasticities. The results are more sensitive to socioeconomic assumptions than to climate change or bioenergy scenarios. When considering a world with higher population and lower economic growth (SSP3), consumption per capita drops on average by 9% for crops and 18% for livestock. The maximum effect of climate change on calorie availability is ?6% at the global level, and the effect of biofuel production on calorie availability is even smaller.     

Evaluating the impact of rising fertilizer prices on crop yields

Because of tensions on fossil energy and phosphorus markets, the rise in fertilizer prices observed during the last decades may continue in the future, putting into question production pathways relying heavily on crop intensification. To evaluate how, in this context, economic choices may alter crop yields, we first construct different fertilizer price scenarios to 2050 based on an econometric relation with oil and gas prices. Other possible scenarios, such as the continuation of historical trends, are also considered. The resulting changes in fertilizer price range between +0.8% and +3.6% per year over the 2005–2050 period. These scenarios are tested in a global land‐use model incorporating an endogenous representation of the land–fertilizer substitution. We find that the supply‐side response to rising fertilizer prices could lower crop yields in 2050 from ?6% to ?13%, with a corresponding increase in global cropland area ranging between 100 and 240 Mha if the demand for food and nonfood products has to be met. The sensitivity of these results is tested with regard to assumptions on food consumption, change in potential yield and nutrient use efficiency.     

Modeling climate change and agriculture: an introduction to the special issue

This issue of Agricultural Economics is a special issue containing articles on model performance in assessing the effects of climate change, bioenergy policy, and socioeconomics on agriculture. The contributions present results from a global economic model intercomparison activity undertaken as part of the AgMIP Project ( www.agmip.org ). The origins of the comparison activities can be traced to a project that was organized by the OECD in late 2010 to compare results from three models. The current phase of the research includes 10 models and was designed in part to support of the IPCC fifth assessment report (AR5). The special issue includes seven peer‐reviewed articles that present thematic results from a range of modeling strategies, with partial and general equilibrium modeling as a high level distinction but a myriad of differences within these two model types. A central common element is harmonization on biophysical effects using crop models and socioeconomic effects using drivers from the Shared Socioeconomic Pathways developed as part of the AR5 process. The special issue provides broad insights into how the modeling communities approached the interactions of climate, socioeconomics, bioenergy policy on agricultural outcomes, including land use, prices, consumption, and production.     

Patterns and processes of pasture to crop conversion in Brazil: Evidence from Mato Grosso State

The rate and location of cropland expansion onto cattle pastures in Brazil could affect global food security, climate change, and economic growth. We combined mapping, statistical modeling, and qualitative methods to investigate patterns and processes of pasture to crop conversion (P2C) in Mato Grosso State (MT), Brazil, a globally important center of agricultural production. P2C land constituted 49% of cropland expansion from 2000 to 2013. For a random sample of ̃250 m pixels in MT, we estimated a regression model skilled at predicting P2C land in the rest of the state as a function of cattle ranching suitability, cropping suitability, and P2C conversion costs. Surprisingly, just 1/7 of pasture agronomically suitable for cultivation had undergone P2C. Hedonic regressions revealed that agronomic characteristics of land were associated with less than 20% of the variation in cropland suitability. Instead, the majority of the variation stemmed from a combination of proximity to agricultural infrastructure, characteristics of neighboring lands, and time fixed effects. The weak relationship between agronomic characteristics of land and P2C location suggests a less certain future for P2C than projections made with agronomic models. Consequentially, complications may arise for greenhouse gas mitigation policies in Brazil predicated on widespread expansion of cropland on pasture vs. natural areas. Our follow-up qualitative research shows that because P2C has often involved land rentals or sales, poorly functioning land institutions may have constrained P2C. Locally poor land quality, omitted from agronomic P2C predictions, can either catalyze or constrain P2C by limiting returns to ranching, farming, or both. Interventions to control rates and locations of P2C should take these insights into account.     

Short‐run and long‐run dynamics of farm land allocation: panel data evidence from Denmark1

This study develops and estimates a dynamic multi‐output model of farmers’ land allocation decisions that allows for the gradual adjustment of allocations that can result from crop rotation practices and quasi‐fixed capital constraints. Estimation is based on micro‐panel data from Danish farmers that include acreage, output, and variable input utilization at the crop level. Results indicate that there are substantial differences between the short‐run and long‐run land allocation behaviour of Danish farmers and that there are substantial differences in the time lags associated with different crops. To our knowledge, this is the first dynamic micro‐model of land allocation estimated on data from the temperate climate zone. Since similar farming conditions are found in northern Europe and parts of the United States and Canada, this result may be of wider interest.     

Agricultural land use change in Karamoja Region,Uganda

We examine dramatic cropland expansion in Karamoja, Uganda by investigating the links between biophysical and political historical events leading to the current state of agricultural land use. Our objective was to quantify agricultural expansion, uncover the dominant drivers leading to the current state of agricultural land use and its impacts on livelihoods. Region wide analysis of remotely sensed data, land use policy and history as well as farmer interviews were undertaken. We posit that government programs instituting sedentary agriculture are the most significant drivers of cropland expansion in Karamoja. We show a 299% increase in cropland area between 2000 and 2011 with most expansion occurring in Moroto District (from 706 ha to 23,328 ha). We found no evidence of an increase in overall crop production or food security and food aid continues to be essential due to recurrent crop failures. Due to lack of resources for inputs (e.g., seeds and labor) cultivated fields remain very small in size and over 55% of once cultivated land is left fallow. Our findings bring into question whether continued promotion of rain-fed agriculture in Karamoja serves the best interests of the people. Current cropland expansion is directly competing and compromising pasture areas critical for livestock-based livelihoods. Without strong agricultural extension programs and major investments in climate-smart options, cropland expansion will continue to have a net negative impact, especially in the context of current climate projections which indicate a future decrease in rainfall, increase in temperature and an increase in the frequency and magnitude of extreme events.     

Addressing climate change cause and effect on land cover and land use in South Asia

This paper evaluates the role of trade liberalization and agricultural intensification in mitigating climate change cause and effects on land use and emissions using a computable general equilibrium model. Our results indicate that cropland expansion triggered by climate-induced crop productivity changes results in deforestation and increases emissions in South Asia and globally. Global full trade liberalization on all goods is the optimum policy for South Asia despite significant global deforestation, but for the world, unilateral partial trade liberalization on all goods is a more appropriate policy while ensuring a considerable emissions reduction for South Asia. These results indicate that mitigation responses to climate change are location specific and no one trade policy is suitable at the regional and global levels. Lastly, agricultural intensification by improving productivity growth is the best strategy in land-based emissions mitigation, thereby avoiding the transformation of forest and pasture lands for agricultural cultivation both at regional and global levels.     

Land use and economic effects of alternative biofuel policies in Brazil and the United States

Being the two largest ethanol producers in the world, biofuel policies in Brazil and the United States affect both their domestic markets and the global food and biofuel economy. In this article we develop a price endogenous mathematical programming model to simulate and analyze the impacts of biofuel mandates and trade distortions on land use, agricultural commodity and transportation fuel markets, and global environment. We find that an 80% increase in total biofuel production from its 103 billion liter baseline level to the mandated 183 billion liter level in 2022 can be achieved with less than 2% increase in total cropland use in both countries. In the United States, this would occur with cellulosic biofuels meeting nearly half of the biofuels consumed and produced largely on cropland pasture and corn ethanol meeting the rest of the mandate and resulting in a 2% increase in corn price. In Brazil, the expansion in sugarcane production would be achieved by reducing land under pasture and a marginal increase in intensification of livestock production. In the aggregate, biofuel policies increase economic surplus in both countries by 1% and redistribute the benefits from agricultural consumers to agricultural producers and the fuel sector. Finally, we also find that full implementation of the mandates in North America, China, and the European Union would reduce the global life‐cycle global greenhouse gas emissions by about 5%.     

Global food demand,productivity growth,and the scarcity of land and water resources: a spatially explicit mathematical programming approach

In the coming decades, an increasing competition for global land and water resources can be expected, due to rising demand for food and bio‐energy production, biodiversity conservation, and changing production conditions due to climate change. The potential of technological change in agriculture to adapt to these trends is subject to considerable uncertainty. In order to simulate these combined effects in a spatially explicit way, we present a model of agricultural production and its impact on the environment (MAgPIE). MAgPIE is a mathematical programming model covering the most important agricultural crop and livestock production types in 10 economic regions worldwide at a spatial resolution of three by three degrees, i.e., approximately 300 by 300 km at the equator. It takes regional economic conditions as well as spatially explicit data on potential crop yields and land and water constraints into account and derives specific land‐use patterns for each grid cell. Shadow prices for binding constraints can be used to valuate resources for which in many places no markets exist, especially irrigation water. In this article, we describe the model structure and validation. We apply the model to possible future scenarios up to 2055 and derive required rates of technological change (i.e., yield increase) in agricultural production in order to meet future food demand.     

Modelling outcome‐related risk in choice experiments

In this study, we introduce information on outcome‐related risk as an additional attribute in a choice model of preferences for a land‐based climate change mitigation project. We provide a comprehensive comparison of different model specifications arising from different behavioural assumptions about the way that respondents process information on outcome‐related risk within the choice task. We find significant differences between several specifications in terms of both model fit and WTP estimates. The behavioural assumptions made when choosing a particular model specification, and reasons that motivate them should be made explicit, and consequences of using different specifications should not be ignored.     

Payments for environmental services to promote “climate‐smart agriculture”? Potential and challenges

Payments for environmental services (PES) have gained wide popularity as approaches to promote environmentally friendly land use or agricultural production practices. Yet academics have also voiced concerns against seeing PES as a panacea. This article discusses whether PES is an appropriate and promising approach to promote so‐called “climate‐smart agriculture” (CSA) practices, which we define as agricultural production practices that contribute to CO₂ emission reductions and/or removals and provide benefits to farmers via increased productivity and profits and reduced vulnerability to climate change. PES appears most promising for the promotion of CSA practices in small‐scale farming contexts with low incomes. Effective design, however, requires solid estimates of cost and benefit flows from CSA adoption over time, accounting for differences in socioeconomic and ecological conditions, and addressing the risk of leakage. Funding for such PES will likely have to come from public sources, and seems most promising where synergies with other objectives such as agricultural development, food security, and climate adaptation or other environmental services exist. The potential of alternative approaches for CSA support such as taxation with rebates for CSA practices, CSA‐related investment support such as microcredits, and hybrid approaches such as conditional microcredit should be further investigated.     

Impacts of increased bioenergy demand on global food markets: an AgMIP economic model intercomparison

Integrated Assessment studies have shown that meeting ambitious greenhouse gas mitigation targets will require substantial amounts of bioenergy as part of the future energy mix. In the course of the Agricultural Model Intercomparison and Improvement Project (AgMIP), five global agro‐economic models were used to analyze a future scenario with global demand for ligno‐cellulosic bioenergy rising to about 100 ExaJoule in 2050. From this exercise a tentative conclusion can be drawn that ambitious climate change mitigation need not drive up global food prices much, if the extra land required for bioenergy production is accessible or if the feedstock, for example, from forests, does not directly compete for agricultural land. Agricultural price effects across models by the year 2050 from high bioenergy demand in an ambitious mitigation scenario appear to be much smaller (+5% average across models) than from direct climate impacts on crop yields in a high‐emission scenario (+25% average across models). However, potential future scarcities of water and nutrients, policy‐induced restrictions on agricultural land expansion, as well as potential welfare losses have not been specifically looked at in this exercise.     

The impacts of biofuels targets on land‐use change and food supply: A global CGE assessment

We analyze the long‐term impacts of large‐scale expansion of biofuels on land‐use change, food supply and prices, and the overall economy in various countries or regions using a multi‐country, multi‐sector global computable general equilibrium model augmented with an explicit land‐use module and detailed biofuel sectors. We find that an expansion of biofuel production to meet the existing or even higher targets in various countries would slightly reduce GDP at the global level but with mixed effects across countries or regions. Significant land re‐allocation would take place with notable decreases in forest and pasture lands in a few countries. The expansion of biofuels would cause a moderate decrease in world food supply and more significant decreases in developing countries like India and Sub‐Saharan Africa. Feedstock commodities (sugar, corn and oil seeds) would experience significant increases in their prices in 2020, but other price changes are small.     

The potential impact of climate change on Taiwan's agriculture

This paper intends to estimate the potential impact of climate change on Taiwan's agricultural sector. Yield response regression models are used to investigate the climate change's impact on 60 crops. A price‐endogenous mathematical programming model is then used to simulate the welfare impacts of yield changes under various climate change scenarios. Results suggest that both warming and climate variations have a significant but non‐monotonic impact on crop yields. Society as a whole would not suffer from warming, but a precipitation increase may be devastating to farmers.     

Scenario simulation of land system change in the Beijing-Tianjin-Hebei region

Urbanization-induced land use problems have been haunting China’s urban agglomerations ever since the beginning of this period of unparalleled economic progress. The Beijing-Tianjin-Hebei (BTH) region is no exception, where coordinated development planning has been implemented by the central government to further resolve attributed problems. Land use simulation models can be used to help governments and planners understand how planning and policies affect the future landscape, by developing sustainable land use strategies which may reasonably balance urbanization and eco-environmental protection. In this paper, we explored the characteristics of historical land use dynamics from 2000 to 2015 in the BTH region and simulated its future land use patterns for 2030 by combining the Dyna-CLUE model with a Markov model to deal with some short comings of existing land use models. The main conclusions are as follows: (1) The figure of merits (FoM) based on the method of three-map comparison reached 85.89 %, which indicated that the simulation model has satisfactory accuracy. (2) Land use structures and spatial patterns differed significantly under business as usual (BAU) scenario, cropland protection (CP) scenario and ecological security (ES) scenario, respectively, owing to the variation of the major objectives designed for different scenarios. (3) By scenario analysis and through tradeoffs, the land use mode under the ecological security scenario might be the optimal solution for future coordinated development in the BTH region. These results will provide theoretical basis and meaningful guidance for regional land optimal allocation.     

Farm Type Effects of an EU‐wide Direct Payment Harmonisation

In this study, we analyse how three scenarios involving different levels of harmonisation of common agricultural policy (CAP) decoupled payments in the EU affect the distribution of farm income across regions and farm types. We use the farm type extension of the common agricultural policy impact (CAPRI) model, which captures farm heterogeneity across the EU. The first scenario (NUTS1) assumes uniform per‐hectare payments at the NUTS1 level. The second scenario (MS‐CONV) equalises the per‐hectare rates inside each Member State (MS) and partially harmonises the single payment scheme (SPS) across MS in line with the 2011 Commission proposal. The third scenario simulates a uniform per‐hectare payment at the EU level. Depending on the implementation of the SPS, the NUTS1 flat rate induces a substantial redistribution of payments across farm types and NUTS2 regions, particularly in regions that apply the historical SPS. The MS‐CONV and EU flat‐rate schemes have more significant impacts at the EU‐wide level. In the EU‐15, almost all farms lose payments from MS‐CONV and EU‐wide flat rates, whereas in the EU‐10, almost all farm types gain from these scenarios. Our conservative estimates indicate that the flat‐rate payments could redistribute up to €8.5 billion. Lower land rental costs partially offset the losses of farm income in the EU‐15 from payment redistribution. Land rents drop for all flat‐rate scenarios across most sectors and farm sizes in the EU‐15. In the less productive new MS, the landowners’ rental income is largely unaffected by the introduction of the flat rate.     

Agent‐based Modelling of Climate Adaptation and Mitigation Options in Agriculture

Computer simulation models can provide valuable insights for climate‐related analysis and help streamline policy interventions for improved adaptation and mitigation in agriculture. Computable general equilibrium (CGE) and partial equilibrium (PE) models are currently being expanded to include land‐use change and energy markets so that the effects of various policy measures on agricultural production can be assessed. Agent‐based modelling (ABM) or multi‐agent systems (MAS) have been suggested as a complementary tool for assessing farmer responses to climate change in agriculture and how these are affected by policies. MAS applied to agricultural systems draw on techniques used for Recursive Farm Programming, but include models of all individual farms, their spatial interactions and the natural environment. In this article, we discuss the specific insights MAS provide for developing robust policies and land‐use strategies in response to climate change. We show that MAS are well‐suited for uncertainty analysis and can thereby complement existing simulation approaches to advance the understanding and implementation of effective climate‐related policies in agriculture.     

Competition for land in the global bioeconomy

The global land use implications of biofuel expansion have received considerable attention in the literature over the past decade. Model‐based estimates of the emissions from cropland expansion have been used to assess the environmental impacts of biofuel policies. And integrated assessment models have estimated the potential for biofuels to contribute to greenhouse gas (GHG) abatement over the coming century. All of these studies feature, explicitly or implicitly, competition between biofuel feed stocks and other land uses. However, the economic mechanisms governing this competition, as well as the contribution of biofuels to global land use change, have not received the close scrutiny that they deserve. The purpose of this article is to offer a deeper look at these factors. We begin with a comparative static analysis which assesses the impact of exogenously specified forecasts of biofuel expansion over the period: 2006–2035. Global land use change is decomposed according to the three key margins of economic response: extensive supply, intensive supply, and demand. Under the International Energy Agency's “New Policies” scenario, biofuels account for nearly one‐fifth of global land use change over the 2006–2035 period. The article also offers a comparative dynamic analysis which determines the optimal path for first and second generation biofuels over the course of the entire 21st century. In the absence of GHG regulation, the welfare‐maximizing path for global land use, in the face of 3% annual growth in oil prices, allocates 225 Mha to biofuel feed stocks by 2100, with the associated biofuels accounting for about 30% of global liquid fuel consumption. This area expansion is somewhat diminished by expected climate change impacts on agriculture, while it is significantly increased by an aggressive GHG emissions target and by advances in conversion efficiency of second generation biofuels.