Evaluating the effects of climate change on US agricultural systems: sensitivity to regional impact and trade expansion scenarios

Agriculture is one of the sectors that is expected to be most significantly impacted by climate change. There has been considerable interest in assessing these impacts and many recent studies investigating agricultural impacts for individual countries and regions using an array of models. However, the great majority of existing studies explore impacts on a country or region of interest without explicitly accounting for impacts on the rest of the world. This approach can bias the results of impact assessments for agriculture given the importance of global trade in this sector. Due to potential impacts on relative competitiveness, international trade, global supply, and prices, the net impacts of climate change on the agricultural sector in each region depend not only on productivity impacts within that region, but on how climate change impacts agricultural productivity throughout the world. In this study, we apply a global model of agriculture and forestry to evaluate climate change impacts on US agriculture with and without accounting for climate change impacts in the rest of the world. In addition, we examine scenarios where trade is expanded to explore the implications for regional allocation of production, trade volumes, and prices. To our knowledge, this is one of the only attempts to explicitly quantify the relative importance of accounting for global climate change when conducting regional assessments of climate change impacts. The results of our analyses reveal substantial differences in estimated impacts on the US agricultural sector when accounting for global impacts vs. US-only impacts, particularly for commodities where the United States has a smaller share of global production. In addition, we find that freer trade can play an important role in helping to buffer regional productivity shocks.

Europe (EUR), the Middle East and North Africa (MNA), and Latin America and the Caribbean 1 (LAC). 2 Finally, grassland yields are projected to increase nearly unanimously across all regions 3 and scenarios, with notable exceptions being Sub-Saharan Africa and Latin America. In general, 4 these results suggest that grassland productivity could offset some of the lost productivity of 5 primary livestock feed grains such as corn. As discussed in this section, these productivity   However, a perfect reproduction is not possible because the trade flows are driven also by other 8 drivers, in particular international trade agreements developments, which are considered constant 9 in the model.

10
There is a large diversity in approaches how global economic models represent the 11 international trade (von Lampe et al. 2014). Most of the partial equilibrium models consider a 12 non-spatial equilibrium, with homogenous goods assumption, and exogenous, fixed, trade cost.

13
On the other hand, the general equilibrium models typically make the assumption of 14 heterogeneous goods, where an Armington elasticity determines the flexibility in substitution 15 between domestically produced and imported products. The world pool market with homogenous 16 goods will typically lead to much more flexibility in international trade adjustments to a climate 17 shock than the Armington spatial equilibrium representation (Stehfest et al. 2013 Table 3 shows livestock commodity impacts, including production and prices. For each 12 region and trade scenario combination, these reported values represent the percentage difference 13 from the no climate change baseline in 2050, averaged over all climate change scenarios. Most 14 livestock commodities summarized here see lower total production and higher prices, with 15 average price impacts for pork, chicken, and eggs exceeding 10% under base trade assumptions.

16
Production and price impacts for milk are smaller but are still negative and positive, respectively.

17
However, beef production expands, and prices are lower, on average, relative to the no climate 18 change baseline across these impacts scenarios. The strong and positive yield effect on US 19 grasslands under the climate change scenarios considered helps maintain productivity of US 20 cattle production systems, which tempers global market and price effects of climate change and 21 reduced feed availability for livestock specifically. In these scenarios, US hog and poultry production systems are more vulnerable to reduced feed grain availability and higher input costs; 1 hence, production declines.   Similar relationships between market power and net differences between global and 9 domestic impacts scenarios shown for crop commodities in Figure 4 in the main body of the 10 paper are seen in projected livestock sector impacts. Results presented in Figure 5 indicate that 11 for livestock commodities, the commodity's share of global exports is a possible indicator of the 12 potential difference between impacts in domestic and global scenarios. US-produced bovine meat shows the largest relative shift in net production and price impacts between the domestic 1 and global scenarios. Although US beef has commanded a significant share of total global 2 production recently (18%), the total export share has been smaller (6%) (FAO 2016). US pork 3 and poultry meat exports command a higher global market share; thus, price and production 4 changes are similar between domestic and global impacts scenarios.

5
However, it is important to note that the US bovine meat system is boosted by climate 6 change productivity shocks on grassland, and this factors into the net change in production and 7 price impacts relative to the no climate change baseline. Under domestic impacts scenarios, US 8 production increases slightly (on average) relative to the no climate change baseline. However, 9 under global scenarios, beef production sees a modest decrease relative to the baseline as less 10 land shifts into grassland in the United States and crop production expands slightly relative to the 11 USA scenarios as the United States seeks to meet global grain demand with reduced productivity 12 in all regions. Thus, while market share likely plays a role in explaining the net change in price 13 and production impacts for US beef when moving from domestic to global impacts scenarios, 14 part of this shift is likely driven by the general supply response present in US beef production 15 systems driven by improved forage productivity. The first robustness check was applied to projected yield changes from crop models.   There are a few key takeaways from our LPJmL scenarios worth noting. First, projected impacts 12 are both less extreme, with smaller productivity and price impacts overall. In many cases, 13 impacts are positive (e.g., increased yields and production) as projected exogenous yields 14 changes higher relative to the no climate change scenario for many crop-region combinations. LPJmL scenarios than for EPIC-derived simulations ( Figure A10 and Figure A11). Most LPJmL 20 scenarios, including RCP 8.5 without CO2 fertilization, show impacts that fall within the mid-21 range of the distribution of impacts from EPIC scenarios, but the direction of these impacts 22 varies in some cases (e.g., less U.S. area devoted to corn production given the modest global Soybean systems see increased yields for almost all region and RCP/GCM combinations 1 under the LPJmL scenarios, unlike EPIC projections, which show mostly negative impacts 2 (especially for soybean producing regions). When comparing domestic and global zone of 3 impact scenarios, we find results that are opposite in sign from the EPIC simulations, but 4 consistent in logic. Higher yields globally under the global zone of impact scenarios relaxes 5 pressure on U.S. soybean systems, which reduces yields and production, with a small (<5%) 6 mean decrease in the price impact as well (averaged across all RCPs and GCMs). Similar results 7 for soybean systems are seen for T2. 8 Crop groups for which the US maintains a smaller total net market share, including 9 wheat, also show net differences across alternative zone of impact and trade scenarios. Wheat 10 shows similar directional changes to soybean systems when comparing domestic and global zone 11 of impact scenarios, though the relative mean impact difference in prices is larger for wheat than 12 for soybean or corn systems, which is consistent with our findings under the EPIC simulations.  The policy implication of this result is that while zone of impact and trade considerations 20 do still matter when projecting impacts from less extreme scenarios, they matter far less than for 21 impact assessments with more pessimistic exogenous productivity shocks. That is, while this 22 manuscript argues that domestic impact assessments which ignore connections to global markets and possible trade adjustments may over-or under-project climate change impacts, the 1 magnitude of this bias is likely smaller when anticipated productivity changes are close to zero 2 or positive. Tariffs   4 The second set of robustness checks were applied to the alternative trade scenario 5 specification. The T2 scenario design was developed to assess possible market outcomes and using the EPIC crop model inputs. We find that reducing existing tariffs has a meaningful effect 4 on net impacts by reducing trade costs, which lowers overall market price impacts and shifts the 5 distribution of other impact metrics. With reduced tariffs, there is less variability in corn 6 production impacts, and price changes are smaller. U.S. soybean area changes relative to the no 7 climate change baseline are larger under T3 than T0, with a smaller decrease in total production 8 as the U.S. expands soybean exports under T3 relative to T0 climate scenarios. Soybean price 9 impacts are also smaller for T3 than for T0 as trade costs are lower. Wheat impacts are also less 10 extreme for T3 than for T0.