4 U.S. Department of Agriculture, Economic Research Service, 1400 ..... (irn), Israel (isr), Kuwait (kwt), Oman (omn), Qatar (qat), Saudi Arabia (sau), United Arab.
Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios Annex: Supplementary Information Keith Wiebe1, Hermann Lotze-Campen2,3, Ronald Sands4, Andrzej Tabeau5, Dominique van der Mensbrugghe6, Anne Biewald2, Benjamin Bodirsky2, Shahnila Islam1, Aikaterini Kavallari7, Daniel MasonD’Croz1, Christoph Müller2, Alexander Popp2, Richard Robertson1, Sherman Robinson1, Hans van Meijl5, Dirk Willenbockel8
Submitted to Environmental Research Letters on 26 November 2014, revised and resubmitted on 24 April 2015 and 15 July 2015, accepted on 19 July 2015.
This Annex describes 1. Model changes since AgMIP Phase 1 2. Intrinsic Productivity Growth Rates (IPRs) 3. Specification of the trade scenarios 4. Additional results
Authors’ affiliations 1
International Food Policy Research Institute, 2033 K St NW, Washington, DC 20006, USA. Potsdam Institute for Climate Impact Research, Telegraphenberg A 31, 14473 Potsdam, Germany. 3 Humboldt University Berlin, Unter den Linden 6, 10099 Berlin, Germany. 4 U.S. Department of Agriculture, Economic Research Service, 1400 Independence Ave. SW, Mail Stop 1800, Washington, DC 20250-0002, USA. 5 LEI Wageningen UR, Droevendaalsesteeg 4, 6708 PB Wageningen, The Netherlands. 6 Global Trade Analysis Project, Purdue University, 403 W State St, West Lafayette, IN 47907, USA. 7 Food and Agriculture Organization of the United Nations, Via delle Terme di Caracalla, Rome 00153, Italy. 8 Institute of Development Studies, University of Sussex, Brighton, East Sussex BN1 9RE, UK. 2
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
1. Model changes since AgMIP Phase 1 The models used in this analysis have been described previously (van der Mensbrugghe 2013, Sands et al. 2014, Woltjer et al. 2014, Rosegrant et al. 2012, Lotze-Campen et al. 2008, Popp et al. 2010, Schmitz et al. 2012, Nelson et al. 2014a and 2014b, von Lampe et al. 2014, Valin et al. 2014, Schmitz et al. 2014, Müller and Robertson 2014). Changes to the models subsequent to the previous work are described below, along with additional information on data and methods.
1.1 The ENVISAGE model ENVISAGE has been updated to GTAP release 8.1 from release 8.0. The impacts globally should be relatively minor. The new release includes 5 new countries in Sub-Saharan Africa and updates to 12 existing input/output tables including some large countries such as Brazil, Republic of Korea, Japan and Nigeria. The more significant change is that a new household demand system has been implemented with a new parameterization (i.e. new income and price elasticities). The old system used the oft-used Linear Expenditure System (LES), with a sliding set of parameters (marginal propensity to consume and floor consumption) that were determined in a base run to achieve targeted trends in budget shares. This generated at times implausible income elasticities and thus rendering comparative static analysis problematic. The new demand system is based on the constant differences in elasticities (CDE) demand function, popularized by the GTAP model. It allows for significantly more flexibility in terms of crossprice elasticities than the LES and suffers less from poor dynamic behavior than the LES, though still far from ideal. Whereas the LES has income elasticities that converge towards unity in the long-run in the absence of adjustments to the parameters, the CDE tends to have relatively flat income elasticity trends, i.e. they stay close to their initial values. Based on work undertaken by the MAGNET team at LEI, a dynamic set of income and price elasticities has been developed for the ENVISAGE model. A calibrated set of CDE parameters is developed at the full GTAP level of disaggregation, i.e. all 57 commodities and all 140 countries/regions in a partial equilibrium framework for each SSP. These calibrated CDE parameters are calculated so as to achieve a targeted set of income elasticities that are generated econometrically using parameters developed for MAGNET. Each SSP, i.e. population/GDP combination, has its own set of calibrated CDE parameters. The latter are then aggregated to the relevant ENVISAGE aggregation for the work undertaken herein. In summary, we have a set of CDE parameters, which, in the absence of any shock to prices, would replicate the targeted income elasticities developed for
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
MAGNET. The actual income elasticities will likely deviate some from the target as prices deviate from their base levels, but these deviations are likely to be small. In any case, the CDE parameters are precalibrated for each of the SSPs (though time dependent), and are also invariant to shocks to the SSPs (for example from climate or trade policies).
1.2 The FARM model The FARM model was updated from GTAP 7 to GTAP 8.2 social accounts, with a 2004 base year. The FARM model operates in five-year steps from 2004 to 2054. Scenarios in FARM use exogenous population time series from SSP1, SSP2, and SSP3. Regional GDP was aligned with SSP1, SSP2, and SSP3 through an iterative process of adjusting labor productivity growth rates for each of 13 FARM regions until GDP in FARM approximates GDP time series from Shared Socio-economic Pathways.
The other major change is that FARM now operates with a full set of 18 Agro-Ecological Zones (AEZs). Previously, 18 AEZs were aggregated to six land classes based on length of growing period. The FARM model still uses a land allocation method that is different from other global economic models in this study. Land is allocated to crops, pasture, and forest by a land market in each AEZ-region combination. This means that land rent per hectare is equal across all uses within an AEZ. The primary motivation for this land use structure is that land values and quantities are preserved for any pattern of land allocation. Further documentation on the FARM model is provided in Sands, Jones, and Marshall (2014).
1.3 The IMPACT model The IMPACT model consists of a suite of linked modules organized around a core global, multi-market, partial equilibrium model of agricultural production, demand, international trade, and prices. The latest version, IMPACT 3, is based on data for 2005 and includes 58 agricultural commodities, 159 countries, 154 water basins, and 320 sub regions (Food Producing Units, or FPUs). Compared to IMPACT 2, which was used in the AgMIP Phase 1 project, IMPACT 3 uses the new disaggregated data base and updated versions of the water models.
The national models include three markets in each country: farm gate (producer prices), national (consumer prices), and international (world prices of exports and imports). Agricultural commodities can be specified as tradable, with prices determined in international markets, or non-traded, with different
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
prices determined in national markets. The model includes value chains, with “activities” producing “commodities” that are used in further processing (e.g., oil seeds to oil and meal or non-traded sugar cane/beet to traded processed sugar).
The IMPACT 3 model system includes a number of linked modules, including: (1) new water models (hydrology, water basin management, and water stress on crops); (2) links to global climate models; (3) DSSAT crop models; (4) a malnutrition module; (5) a welfare analysis module; (6) spatial production allocation models (SPAM); (7) a livestock model (feed grains to livestock to processed meat); and (8) a land-use model that determines the allocation of crops to irrigated and non-irrigated land by FPU, with an endogenous shadow price of land equilibrating the supply of and demand for land by type. New modules on land use, livestock, fisheries, and commodity demand that account for nutrition are under development.
1.4 The MAGNET model LEI moves from previously used LEITAP/MAGNET Model to the flexible MAGNET model. A distinguishing feature of the model is its modular structure, which allows the model structure to be tailored to the research question at hand. Also, it allows for flexible aggregation of sectors and regions depending on the particular research requirements.
Compared with LEITAP/MAGNET version, several important improvements are included. On the demand side, a module calculating the nutrient content consumption of the private household is included. Next, the private household demand system is recalibrated under the assumption that the total calories consumption should not differ significantly from expected by FAO-AT report.
On the supply side, lower estimates of maximum available agricultural land obtained by Netherlands Environmental Assessment Agency (PBL) are used. Also, based on literature search and land use time series data analysis, land supply function elasticities in respect of land price are recalibrated. Lower elasticities values for these are obtained and used compared with previous version of the model. This leads to lower agricultural land area responses and higher agricultural land price responses to changes of demand of agri-food products, compared with previous version of the model.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
The new version of the model represents explicitly thee fertilizer types - N, P and K - and assumes Leontief substitution between them. The substitution elasticity between agricultural land and fertilizer used in the production process depends on development level of the country or region. For low developed regions, the elasticity equal 0.5 is assumed. It decreases gradually when region becomes richer and drops to 0.15 for high developed regions.
While LEITAP/MAGNET was using GTAP 6 database and 2001 base year, MAGNET is now using GTAP 8.1 data and the base year of the model is 2007.
1.5 The MAgPIE model The most important improvement has been the implementation of flexible pasture area in the model, which in the previous version was static. The new implementation includes a dynamic demand for pasture products, flexible pasture area and the option to increase the intensity of pasture areas. Depending on regional feed baskets, ruminant production creates a demand for a certain quantity of grass. This demand has to be settled by pasture land, for which potential yields are derived with the LPJmL model. An increase in production can be reached either through investment into research and development or by increasing the pasture area. As pasture area is now dynamic and competes with other land-uses, the new implementation creates a more realistic representation of land scarcity and therefore improves the estimation of land-use patterns and food prices.
In the new version of MAgPIE, it is now also possible to increase the area equipped for irrigation, meaning the model can endogenously invest into irrigation infrastructure. This leads to a higher flexibility in investment decisions and a more efficient use of scarce water resources.
Input data from LPJmL for the potential yields and water availability representing the climate scenarios are now bias corrected and diverge after 2005.
All the SSP scenarios have been further developed and adjusted to the storylines as well as GDP and population input data. The impact of future changes in animal-based calories in diets and overall food demand on the agricultural system has been implemented through exogenous income and time dependent scenarios, derived from historical data (Bodirsky et al. (submitted), Valin et al. 2014). The
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
resulting food demand is lowest in SSP 1 and highest in SSP 3. Because of lower income levels, the demand for livestock products is lowest in SSP 3, followed by SSP 1 where livestock consumption is low due to more sustainable dietary preferences. Protected land areas (e.g. tropical forests) increase by 120%, 20% and 0% until 2050 relative to 2010 for SSP 1, SSP 2 and SSP 3 respectively. The increases in livestock productivity for the SSP scenarios are higher than in AgMIP Phase 1. In SSP 1 and SSP 2, about two thirds of global livestock in 2050 is produced at European productivity levels; in SSP 3 this applies to only about one third. The latest version of the SSP scenario parameterization improves the comparability across models and with the upcoming official SSP scenarios.
1.6 The LPJmL model The LPJmL model version used here is the same as that used in AgMIP Phase 1 (see Müller and Robertson 2014).
2. Intrinsic Productivity Growth Rates (IPRs) Technological change will have a major effect over time on the capacity of production, and must be considered one of the major drivers of economic change in the future. In the agriculture sector this is especially true as critical factors of production in many regions of the world are already saturated for their agricultural use, and the ability to produce more from this scarce resource is dependent on new technologies allowing more efficient and productive uses of this scarce resource. Technological change is driven by a variety of public and private responses to market and non-market forces. To better interpret ex-ante changes in productivity it is important to isolate the different responses to ensure that behavior is correctly simulated. To this end, IFPRI has developed a series of commodity- and countryspecific assumptions on non-price (exogenous) productivity growth (through 2050) for the IMPACT model (International Model for Policy Analysis of Agricultural Commodities and Trade). These exogenous assumptions on agricultural productivity are called Intrinsic Productivity Growth Rates (IPRs). (The private farmer response to market forces (prices) is captured in the endogenous area and yield response functions inside of IMPACT.) The IPRs attempt to summarize historical trends on productivity increases, and expert opinion on the future returns of agriculture R&D. They summarize the improvements that
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
can be achieved from a variety of technological improvements in the agriculture sector through advances in:
Management Practices (i.e. tilling, crop spacing/intensification, laser-land leveling, etc.)
Plant Breeding (conventional, and wide-crossing-hybridization)
Biotechnology (GMOs, transgenic, etc.)
Agricultural Extension and improving the diffusion of knowledge across the Ag-sector
The IPRs defined in 5-year periods, allowing for a transition from historical trends in the early part of the projection period and generally slowing over time reflecting expert opinion on biological limits to continued productivity growth and diminishing returns to research. The IPRs were developed as a baseline assumption on productivity growth through 2050, as a starting point or counterfactual for scenario analysis on changes in productivity growth. They have been estimated to correspond with a socio-economic future that generally follows recent historical trends, which corresponds with the IPCC’s second Shared Socioeconomic Pathway (SSP2). It would be expected that changes in the assumptions on economic growth would have consequences towards to potential productivity gains that could be achieved. Slower economic growth all things equal will lead to less resources for Agricultural R&D, which would lead to slower productivity growth. As efforts are made to develop Representative Agricultural Pathways (RAPs) to describe the functioning of the Agriculture Sector with in the SSP framework, it has become evident that the IPRs need to be adjusted to reflect the availability of resources for investment in Agricultural R&D. To create this adjustment the current IFPRI IPRs were assigned to the SSP2 scenario, and differences in economic growth between each SSP and SSP2 were used to adjust downward or upward the productivity trends. While we could have used changes in income (per capita GDP growth), GDP growth was chosen as the primary driver to adjust the IPRs due to the critical role of public investment in agricultural R&D, especially in developing countries, and changes in GDP reflect the potential resources available for public investment.
GDP growth across the SSP scenarios varies greatly by region, and it is clear that certain regions due to current low productivity have the potential for more rapid productivity growth (yield gaps). To reflect this reality the adjustments to the IPRs also reflect that different regions should have different productivity responses to changes in economic growth. As a first attempt at capturing these differences the world was divided into 3 groups of countries each with a different productivity response to changes
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
in GDP growth. The 3 groups used were developed countries (DVED)1, BRIC2 countries, and the other developing countries (DVNG). The sensitivity of the adjustment reflected expert opinion on the relative role of economic growth on changes in agricultural productivity, and more or less reflect that global average on the relationship between TFP growth and GDP growth.
IPR Adjustment The adjustment of the IPRs in this first attempt at developing RAPs for SSPs was primarily mechanistic, which the following equations will explain. First, a national adjustment factor was calculated for each SSP due to the deviation of GDP growth rate over each 5-year period from the SSP2 baseline.3
Equation 1 Calculating National IPR adjustment factors by SSP 𝐼𝑃𝑅𝐴𝑑𝑗𝑢𝑠𝑡𝑆𝑆𝑃𝑋,𝑐𝑡𝑦,𝑝𝑒𝑟 = 𝐺𝐷𝑃𝑆𝑛𝑠𝑐𝑡𝑦𝑔𝑟𝑜𝑢𝑝 × (𝐺𝐷𝑃𝑔𝑟𝑆𝑆𝑃2,𝑐𝑡𝑦,𝑝𝑒𝑟 − 𝐺𝐷𝑃𝑔𝑟𝑆𝑆𝑃𝑋,𝑐𝑡𝑦,𝑝𝑒𝑟 ) GDPSns IPR sensitivity to changes in GDP growth (DVED 0.05, BRIC 0.10, DVNG 0.20) GDPgr Annual GDP growth rate cty Country per 5-yer time period Once the national adjustment factor have been calculated it needs to be scaled to the commodity specific IPR to ensure we do not generate unrealistic changes in productivity growth assumptions. This was done in two ways. First, the livestock sector adjustment was scaled to take into account the slower growth of livestock productivity vis-à-vis crop productivity.
Equation 2 Scaling national IPR adjustment factor for livestock sector 𝐼𝑃𝑅𝐽𝐴𝑑𝑗𝑢𝑠𝑡𝑆𝑆𝑃𝑋,𝑗,𝑐𝑡𝑦,𝑝𝑒𝑟 = 𝐼𝑃𝑅𝐴𝑑𝑗𝑢𝑠𝑡𝑆𝑆𝑃𝑋,𝑐𝑡𝑦,𝑝𝑒𝑟 × 𝑆𝑐𝑎𝑙𝑒𝑃𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟𝑗 IPRJAdjust Commodity specific IPR adjustment IPRAdjust National level IPR adjustment ScaleParameter Scaling parameter to adjust national level IPR to commodity Meat producing animals 0.5, milk and egg producing animals: 0.1 j IMPACT Commodity cty Country per 5-yer time period
1
Australia, Canada, EU 27, Israel, Japan, New Zealand, Norway, South Korea, USA Brazil, Russia, India, and China. South Africa and Indonesia were left with the other developing countries 3 It should be noted that the IPR Adjustment for SSP2 will be 0, as we’ve assigned the base IPRs to the SSP scenario 2
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Finally, a last data check was done to ensure unrealistic increases (or decreases) in productivity growth are not applied in the scenarios. This final check involved implementing a ceiling and a floor on the allowed increase (or decrease) of productivity in the scenario vis-à-vis the SSP2 baseline. The floor that was chosen was 20 percent of the original SSP2 IPR to ensure that we don’t create negative yield growth. The ceiling caps the IPR adjustment at 3 times the original SSP2 scenario IPR. Once these final checks have been applied to the IPR commodity specific adjustment we calculate each SSP-specific IPR in the following way:
Equation 3 Final Calculations for SSP specific IPRs 𝐼𝑃𝑅𝑆𝑆𝑃𝑋,𝑗,𝑐𝑡𝑦,𝑝𝑒𝑟 = 𝐼𝑃𝑅𝑆𝑆𝑃2,𝑗,𝑐𝑡𝑦,𝑝𝑒𝑟 − 𝐼𝑃𝑅𝐽𝐴𝑑𝑗𝑢𝑠𝑡𝑆𝑆𝑃𝑋,𝑗,𝑐𝑡𝑦,𝑝𝑒𝑟
3. Specification of the trade scenarios Trade scenario description The following are the trade scenarios implemented in this paper.
For SSP2 no change to trade policies.
SSP1 is a more globalized world. We remove all tariffs and export subsidies on all trade in agricultural and food products. The trade measures are phased out over the period 2020 through 2035, i.e. trade policies in 2020 are still the baseline policies with liberalization starting in 2021 and with all measures 0 in the year 2035. Domestic support policies are not being affected.
The policy formula is therefore the following:
if 0 t 2020 t 0 f 0 if 2035 2020 if f
t 2020 2020 t 2035 t 2035
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
where 0 represents the baseline trade measure and f the final level (in the case of free trade equal to 0).
The strategy for SSP3 is more complex. The idea is to divide the world into three trade blocks: East and South Asia, a broad European block that encompasses Western and Eastern Europe, Central Asia, the Middle East and Africa, and a Western Hemisphere block that include North, Central and South America. Import tariffs between the blocks are doubled (no change to export measures), and no change to trade policies within blocks. There are a number of ways to double tariffs (at constant trade shares). We explore three options. 1. The bilateral tariff between blocks is doubled. Let r represent the base year tariff imposed on block r 0
(where r is either 1 or 2 for expositional purposes), and Tr represents the trade value of imports of block r. Then in this formula, we have:
r1 2 r0 and obviously, the average tariff doubles (with constant trade shares).
2. The average bilateral tariff with all blocks (outside of the own-block) is doubled. In this case we have the following formulas:
a0
10T1 20T2 T1 T2
10T1 20T2 T
r1 r0 a0 Thus the average tariff over the trading blocks is added to the initial tariff of the individual blocks. This has the same average impact, i.e. a doubling of tariffs on trade with the blocks, but the difference in tariffs (in percentage terms) is uniform across all blocks.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
3. The third option is to increase tariffs across the blocks, but maintain the same price ratios, i.e. at constant prices, the trade shares between the 2 trading blocks would be invariant. This is captured with the following constraints:
a1 2 a0 1 10 1 11 1 20 1 21
The solution to this is given by the following formula:
r1
r0 1 2 a0 a0 1 a0
Trade data
Table 1 provides the 2007 value of food imports for the three blocks (with the importers on the top row). Total food trade is about $1 trillion ($US2007), of which a significant share, some $0.43 trillion is composed of trade within the European block (not just the EU). In general the diagonal elements tend to have the greatest weight for each of the three blocks, though in the case of the Asia there is somewhat greater balance. The Western Hemisphere is a net exporter of food, with a net balance of some $100 billion, with roughly 25% accounted for by Asia and the rest with the European block.
Table 1: Food imports for 2007 by the three blocks (importers in columns) ($2007 million)
Exporter\importer Asia block European block Western Hemisphere block Total Source: GTAP v8.1, regions defined in Annex.
Asia
Europe+
West. Hem.
Total
104,966 46,139 77,653 228,757
65,304 428,017 89,470 582,791
30,296 32,625 120,794 183,714
200,566 506,780 287,917 995,262
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Table 2 provides the bilateral average tariffs on food imports across the three broad blocks. The European and Western Hemisphere block have on average the lowest tariffs at around 5 percent, with a much larger 14 percent for the Asian block. However, the European block exhibits much higher tariffs when the intra-regional trade is excluded as the average tariff more than double doubles in this case, with a very low intra-regional tariff of only around 3 percent (much of it accounted for by the European Union). The Asian and Western Hemisphere blocks exhibit less dispersion in tariffs across all three blocks.
Table 2: Tariffs on food imports in 2007 (importers in columns) (percent)
Asia block European block Western Hemisphere block Total Extra-block
Asia 15.2 16.7 11.1 14.1 13.2
Europe+ 10.5 2.9 11.2 5.0 10.9
West. Hem. 5.4 6.9 4.9 5.3 6.2
Total 12.2 4.4 8.5 7.1
Source: GTAP v8.1, regions defined in Annex.
Table 3 shows the patterns of tariffs under the three different scenarios proposed above. The top panel corresponds to the first scenario above. The tariffs of each of the extra-blocks are doubled. The second panel represents the second option—tariffs of each of the extra blocks are incremented by the same percentage amount, consistent with a doubling of the extra-block average tariff. Thus for the Asian block, European block tariffs would increase to 30 percent instead of 33.5 percent if they are simply doubled. For the Western Hemisphere block the impact is reversed, with a somewhat higher tariff than in the case of a simple doubling of the block’s initial tariff rate. The third panel shows the level of tariffs if relative tariff-inclusive prices are held fixed. The impacts are relatively close to the second panel.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Table 3: Tariff scenarios (importers in columns) (percent)
2 x block tariff Asia block European block Western Hemisphere block Average Extra-block
Asia
Europe+
West. Hem.
15.2% 33.5% 22.3% 21.3% 26.5%
20.9% 2.9% 22.4% 7.9% 21.8%
10.8% 13.7% 4.9% 7.4% 12.3%
Asia 15.2% 30.0% 24.4% 21.3% 26.5%
Europe+ 21.3% 2.9% 22.1% 7.9% 21.8%
West. Hem. 11.5% 13.0% 4.9% 7.4% 12.3%
Asia 15.2% 30.4% 24.1% 21.3% 26.5%
Europe+ 21.3% 2.9% 22.1% 7.9% 21.8%
West. Hem. 11.5% 13.1% 4.9% 7.4% 12.3%
2 x avg. tariff extra block Asia block European block Western Hemisphere block Average Extra-block Uniform price ratios Asia block European block Western Hemisphere block Average Extra-block
Given the simplicity of the second scenario, we chose to add 13.23 percent to the tariffs of Asian block trade partners, 10.89 percent to the tariffs of the European block trade partners and 6.16 percent to the trade partners of the Western Hemisphere block.
Annex: Definition of regions and sectors for ENVISAGE Regions The following table shows the regional aggregation based on V8.1 of the GTAP database. Note that the ‘xhy’ region includes some European countries.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Asian block anz
Australia & New Zealand Australia (aus), New Zealand (nzl)
jpn xhy
Japan Rest of high-income Hong Kong (hkg), Korea (kor), Taiwan (twn), Switzerland (che), Norway (nor), Rest of EFTA (xef), Rest of the World (xtw)
chn idn xea
China Indonesia Rest of East Asia Rest of Oceania (xoc), Mongolia (mng), Rest of East Asia (xea), Cambodia (khm), Laos (lao), Malaysia (mys), Philippines (phl), Singapore (sgp), Thailand (tha), Viet Nam (vnm), Rest of Southeast Asia (xse)
ind xsa
India Rest of South Asia Bangladesh (bgd), Nepal (npl), Pakistan (pak), Sri Lanka (lka), Rest of South Asia (xsa)
European block eur European Union 27 Austria (aut), Belgium (bel), Cyprus (cyp), Czech Republic (cze), Denmark (dnk), Estonia (est), Finland (fin), France (fra), Germany (deu), Greece (grc), Hungary (hun), Ireland (irl), Italy (ita), Latvia (lva), Lithuania (ltu), Luxembourg (lux), Malta (mlt), Netherlands (nld), Poland (pol), Portugal (prt), Slovakia (svk), Slovenia (svn), Spain (esp), Sweden (swe), United Kingdom (gbr), Bulgaria (bgr), Romania (rou)
rus tur xec
Russia Turkey Rest of Europe & Central Asia Albania (alb), Belarus (blr), Croatia (hrv), Ukraine (ukr), Rest of Eastern Europe (xee), Rest of Europe (xer), Kazakhstan (kaz), Kyrgystan (kgz), Rest of Former Soviet Union (xsu), Armenia (arm), Azerbaijan (aze), Georgia (geo)
mna
Middle East & North Africa Bahrain (bhr), Iran (irn), Israel (isr), Kuwait (kwt), Oman (omn), Qatar (qat), Saudi Arabia (sau), United Arab Emirates (are), Rest of Western Asia (xws), Egypt (egy), Morocco (mar), Tunisia (tun), Rest of North Africa (xnf)
ssa
Sub-Saharan Africa Benin (ben), Burkina Faso (bfa), Cameroon (cmr), Côte d’Ivoire (civ), Ghana (gha), Guinea (gin), Nigeria (nga), Senegal (sen), Togo (tgo), Rest of Western Africa (xwf), Central Africa (xcf), South-Central Africa (xac), Ethiopia (eth), Kenya (ken), Madagascar (mdg), Malawi (mwi), Mauritius (mus), Mozambique (moz), Rwanda (rwa), Tanzania (tza), Uganda (uga), Zambia (zmb), Zimbabwe (zwe), Rest of Eastern Africa (xec), Botswana (bwa), Namibia (nam), South Africa (zaf), Rest of South African Customs Union (xsc)
Western Hemisphere arg Argentina bra Brazil mex Mexico xlc Rest of Latin America & Caribbean Rest of North America (xna), Bolivia (bol), Chile (chl), Colombia (col), Ecuador (ecu), Paraguay (pry), Peru (per), Uruguay (ury), Venezuela (ven), Rest of South America (xsm), Costa Rica (cri), Guatemala (gtm), Honduras (hnd), Nicaragua (nic), Panama (pan), El Salvador (slv), Rest of Central America (xca), Caribbean (xcb)
can usa
Canada United States
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Sectors The following table shows the sectoral definitions, of which there are 15 agriculture and food sectors. Note that the fisheries sector has been aggregated with other food and forestry is not considered an agricultural sector.
Food and agriculture ric
Rice Paddy rice (pdr), Processed rice (pcr)
wht gro v_f osd sug
Wheat Other grains Vegetables & fruits Oil seeds Sugar Sugar cane and sugar beet (c_b), Sugar (sgr)
ocr
Other crops Plant-based fibers (pfb), Crops, n.e.s. (ocr)
ctl
Cattle Bovine cattle, sheep and goats, horses (ctl), Wool, silk-worm cocoons (wol)
oap rmk cmt omt vol mil ofd
Other livestock Raw milk Red meat Other meat Vegetable oils Dairy products Other food Fishing (fsh), Food products n.e.s. (ofd), Beverages and tobacco products (b_t)
Natural resources, manufacturing and services frs Forestry coa Coal oil Oil gas Natural gas omn Other mining twp Textile wearing apparel & leather goods Textiles (tex), Wearing apparel (wap), Leather products (lea)
ke5
Energy intensive manufacturing Paper products, publishing (ppp), Chemical, rubber, plastic products (crp), Mineral products n.e.s. (nmm), Ferrous metals (i_s), Metals n.e.s. (nfm)
xmn
Other manufacturing Wood products (lum), Metal products (fmp), Motor vehicles and parts (mvh), Transport equipment n.e.s. (otn),
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Electronic equipment (ele), Machinery and equipment n.e.s. (ome), Manufactures n.e.s. (omf)
p_c ely gdt cns srv
Refined oil products Electricity Gas distribution Construction Services Water (wtr), Trade (trd), Transport n.e.s. (otp), Sea transport (wtp), Air transport (atp), Communication (cmn), Financial services n.e.s. (ofi), Insurance (isr), Business services n.e.s. (obs), Recreation and other services (ros), Public administration and defence, education, health services (osg), Dwellings (dwe)
Implementation of trade scenarios in MAgPIE
In MAgPIE, trade dynamics are estimated with a different approach than in the other GCE and PE models. MAgPIE has two trade pools: the free trade pool, where goods are fully traded according to comparative advantage, and the historical trade pool, where exports and imports stick to historical patterns. The fraction of trade which belongs to the historical trade pool is a scenario parameter. In SSP 1, where trade is strongly liberalized, we assume that the historical trade pool and thus trade barriers are reduced by 10% per decade. With the resulting incomplete liberalization we take into account that regions have non-monetary trade barriers such as preferences for local products. In SSP 2, trade barriers are reduced by only 5% per decade, while in SSP 3 they remain at the level of 2005. Due to our implementation MAgPIE cannot distinguish between import tariffs and export subsidies.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
4. Additional results Figure S1 – Exogenous impacts of climate change on crop yields, by region, for RCP 4.5, 6.0 and 8.5 (% change relative to respective baseline values in 2050 without climate change)
Note: Based on the three selected GCMs as represented by the LPJmL crop model (n = 15). Each dot depicts the result for one crop and one GCM. Crops: WHT = wheat, RIC = rice, CGR = coarse grains, OSD = oilseeds, SUG = sugar. Regions: ANZ = Australia/New Zealand, BRA = Brazil, CAN = Canada, CHN = China, EUR = Europe, FSU = Former Soviet Union, IND = India, MEN = Middle East/North Africa, OAS = Other Asia, OSA = Other Latin America, SEA = Southeast Asia, SSA = Sub-Saharan Africa, USA = United States of America. Source: The authors
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Figure S2 -- Baseline increases in global yields, area, production, consumption, exports, imports and prices of the five commodities in 2050, by economic model (% change relative to 2005 values)
Note: The plots show baseline results for the five commodities for each of five economic models and eight variables, aggregated across thirteen regions. Each dot depicts the result for one crop and one SSP. MAgPIE estimates net trade flows but not exports and imports separately. Source: The authors.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
The model results are broadly consistent with those from AgMIP Phase 1 (Nelson et al. 2014), but also reflect changes to the SSPs (moving from version 0.5 to 0.93) and model changes. Global GDP in SSP 2 is down some 6 percent in 2050 across the two SSP versions. The main declines in GDP are in countries that still have relatively elastic food demand, for example -23 percent in China, -28 percent in India, and -18 percent in Sub-Saharan Africa. The exogenous yield is a key assumption within this modelling exercise and this shock is somewhat lower in the new SSP 2 assumptions (reflecting in part the decline in GDP trends). In the new assumptions it is highest for coarse grains and wheat, followed by sugar, oilseeds and rice. MAgPIE has no exogenous yield assumption as technological change is endogenous in this model. The total yield effect is highest in MAGNET as it has the highest endogenous yield effect. The increased demand in this model leads mainly to additional yield increase and not area increase, as is the case for the FARM model. For ENVISAGE and IMPACT the area increase is much higher and the endogenous yield effect much smaller. For MAGNET this area/yield response is very different from AGMIP phase 1 as the land module is totally recalibrated given new and much lower estimations of the land that is potentially available and suitable for agriculture. For the other models are the differences with AGMIP phase 1 are much less pronounced. Production growth is highest for oilseeds driven by a high income elasticity (e.g. ENVISAGE) and growth for biofuels due to renewable energy policies (e.g. MAGNET). For rice production growth is generally low due to a low increase in demand (low income elasticity, see especially IMPACT results). The main change to ENVISAGE relative to AGMIP Phase 1 entailed a sharp downward revision in income elasticities for agriculture and food—thus weakening the link between income growth and food demand growth. The ENVISAGE global production and consumption increases are down (comparing only the SSP 2 scenarios with no climate change)—an increase of some 58 percent compared with an increase of 77 percent in the Phase 1 results due to the lower GDP assumptions and especially these lower income elasticities. Trade effects are highest for oilseeds, especially in the ENVISAGE and FARM models. Price results are very low in ENVISAGE and MAgPIE indicating that for the future in these three scenarios no large increase in agricultural prices is expected. For MAgPIE this is mainly explained by more flexibility in grassland use, compared to AgMIP Phase 1, leading to higher cropland availability and reduced costs of production. FARM and MAGNET show a clear higher increase in prices in the SSP 3 scenario as the lower GDP assumption in this scenario induces less technical change in these models and the higher population growth leads to a higher demand (see also Robinson et al. 2014). For the IMPACT model the lower GDP in SSP 3 has no negative impact on technical change and here the price developments in SSP 3 are not much higher than in SSP 2.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Figure S3 – Impacts of climate change on global yields, area, production and prices of the 5commodity aggregate relative to baseline values in 2050 for each SSP, compared across the five models
Note: ENV = ENVISAGE, FAR = FARM, IMP = IMPACT, MGN = MAGNET, MGP = MAgPIE. The plots represent individual results for the five economic models, aggregated across the five commodities and thirteen regions. Variables: YEXO = exogenous yield shocks, YTOT = realized yields after management adaptation, AREA = agricultural area in production, PROD = total production, PRICE = price. Source: The authors.
Wiebe et al., Climate change impacts on agriculture in 2050 under a range of plausible socioeconomic and emissions scenarios, Supplementary Information (Environmental Research Letters, accepted 19 July 2015)
Figure S4 – Impacts of climate change on global yields, area, production and prices of the 5commodity aggregate relative to baseline values in 2050 for each SSP, compared across regions
Note: NAM = North America, OAM = Other America, SAS = South Asia, SEA = Southeast Asia, SSA = Sub-Saharan Africa, WLD = World. The plots represent individual results for the five regions, aggregated across the five commodities and five models. Variables: YEXO = exogenous yield shocks, YTOT = realized yields after management adaptation, AREA = agricultural area in production, PROD = total production, PRICE = price. Source: The authors.