Economic impacts of climate change on EU agriculture: will the farmers benefit from global climate change?

This paper analyses how climate change might impact EU agricultural markets by mid-century, considering a large ensemble of climate change projections from different models, market adjustments and trade feedbacks. Applying consistent climate change driven productivity shocks to a global multi-commodity agricultural market model we show that the negative direct effects from climate change on crop production in the EU could be offset by market and trade adjustments. The simulations reveal that climate change has heterogeneous impacts across regions. EU farming sector, in particular, might actually benefit from climate change as the impacts on agricultural productivity are expected to be more severe in important non-EU production regions such as US, Russia and Ukraine, depending on the crop. Higher producer prices for important crops such as wheat, barley, grain maize, rice and soybeans, lead to an increase in EU production and exports. For instance EU wheat production could increase by 13% and exports by 28%, with 19% higher farm incomes on average than in a business-as-usual situation. Our study has several limitations. In particular, we do not consider CO2 fertilization effects and direct effect from climate change on livestock sector, climate variability and extreme weather effects. Notwithstanding, our findings highlight the heterogeneity of climate change impacts across regions, specifically Northern versus Southern Europe, and the importance of market and trade adjustments as economic adaptation mechanisms to climate change.


Introduction
Agricultural production and the farming sector are affected by many different drivers of change, ranging from biophysical, policy, socio-economic and market related ones.In recent decades climate change has become one of the main drivers and there are growing concerns of how this might change the future development of agri-food systems (Von Lampe et al 2014), especially due to prevailing adverse biophysical effects and their impacts on crop yields, as a result of temperature and precipitation changes (IPCC 2014(IPCC , 2022)).These biophysical effects will affect global crop production, prices and market adjustments.Agriculture in the EU will also be affected (Feyen et al 2020), but the projected climate change impacts on agricultural production are wide-ranging (Olesen et al 2011).For instance, on the one side, some Northern European regions could benefit given the higher solar radiation and milder temperatures projections due to climate change allowing for a longer cropping season.On the other side, some Southern European rain-fed crop production regions will likely face considerable difficulties due to higher temperatures and less precipitation.Market feedback can be an important economic adaptation measure to climate change impacts on agriculture (Nelson et al 2014).In fact, a decrease in production in a certain region motivated by a change in climate conditions will exert some pressure on prices, triggering other less affected neighbouring regions to increase production and, therefore, allow for a readjustment of the affected markets.However, there are many uncertainties in understanding climate change impacts on EU agriculture and how the impacts in non-EU production regions could affect European markets through trade adjustments (Jacobs et al 2019).There are wide ranging crop model responses to climate change and isolated from agro-economic responses (Müller et al 2017), thus making the evidence on market and trade feedbacks still unclear.
The existing literature indicates that the economic impacts of climate change on food markets can only appropriately be assessed when taking into account explicitly the biophysical effects on crop yields across the globe in combination with the economic response of global agricultural markets.Although such combined studies exist at the global level considering macro regions (e.g.Nelson et al 2014, Van Meijl et al 2018, Dumortier et al 2021), respective comprehensive regional analysis is rather scarce or non-existent in the context of a large ensemble of climate change scenarios from different climate and crop models globally applied in a consistent manner.Therefore, the aim of this paper is to quantify the possible consequences of climate change for EU agriculture in 2050, considering both biophysical impacts and agricultural market interactions within a consistent modelling framework.More specifically, this paper aims to address the following two major questions: Are market feedback loops important to be considered when analysing climate change impacts on agricultural production at regional level?What is the range of possible consequences of climate change for EU agriculture, i.e. are there regions where EU farming sector is likely to benefit from global climate change and others where they are likely to be worse off?By answering these questions, our results contribute to the better understanding of climate change impacts on agriculture and help fostering a better design of policy measures for the adaptation to climate change, especially as they point out the most vulnerable regions in the EU.

Methodology
To reflect the uncertainty related to climate change in the long term, we use biophysical simulation results provided by the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) Fast-Track simulation database (www.isimip.org/)and take into account the interactions between five General circulation models (GCMs)1 and seven Global Gridded Crop Growth Models (GGCMs).2To capture the feedback of agricultural markets, we introduce the biophysical crop yield effects estimated by these model ensembles in the global multi-commodity agricultural market model Common Agricultural Policy Regional Impact Analysis (CAPRI)3 .CAPRI is then used to analyse in a comparative static framework the impacts on EU agriculture.CAPRI captures agricultural production highly detailed and disaggregated at the EU regional level, following profit maximising behaviour under constraints, such as land availability, nutrient balances and policy obligations.The land supply is a function of the returns to land, i.e. maximal amount of economically usable agricultural area in a region given endogenously calculated land rents.The EU supply interacts with the global market model where bilateral trade and prices for major agricultural commodities are computed considering human and feed consumption, biofuel use and import demand from multilateral trade relations (Britz and Witzke 2014).In addition, the behavioural functions in the market model represent supply for primary agricultural and processed commodities, balancing constraints and agricultural market policy instruments (that is, import tariffs, tariff rate quotas, producer and consumer support estimates, etc).Hence, the impact of exogenous (climate change-related) yield shocks on specific crop production regions (as estimated by the biophysical models) is analysed under explicit consideration of global market adjustments affecting both producer and consumer prices.The endogenous reaction in the model mainly depends on the price adjustment mechanisms (both domestic and international prices), i.e. adjust the production depending on the price elasticities estimated by Jansson and Heckelei (2011).Accordingly, CAPRI is suitable for a comprehensive analysis of climate change effects considering market adjustments and was used in several multimodel assessments at global level (Hasegawa et al 2018, Van Meijl et al 2018), and has also been used before to assess the impacts of climate change on EU agriculture (Shrestha et al 2013, Delincé et al 2015, Blanco et al 2017, Martinez et al 2017).However, in the previous studies focusing on the EU, projections for non-EU regions were mainly based only on a single crop growth model and the analysis was primarily done at EU aggregated level and not considering regional aspects (for instance, differentiating between Northern and Southern European impacts).Moreover, some of the previous EU studies did not consider climate change impacts in non-EU regions and/or market adjustments, which are important adaptation mechanisms.
The ISIMIP Fast-Track simulation database provides information on biophysical yield changes per region and crop from 2011 to 2100.These changes are compared to the average yield during a 30 year historic period  and consider interactions between five GCMs and seven GGCMs.The regional crop-specific yields in ISI-MIP account for water availability limitations and differentiate between irrigated and rain-fed agricultural production.However, in the CAPRI model version used for this project, yields are not separated between rain-fed and irrigated production systems.Therefore, the ISI-MIP data had to be combined, considering both irrigated and rain-fed yields weighted by their respective cultivation areas.Additionally, it is worth noting that the agricultural management options in the GGCMs differ from those in CAPRI.Some GGCMs use fixed management decisions based on historical averages for all future years, while CAPRI considers input prices or yield changes due to climate change for decision-making (table 1).
Given the combination of the GCM and GGCM models, we simulated 35 yield shock scenarios focusing on yield changes in the year 2050 for five crops: wheat, barley, grain maize, rice, and soybeans (see appendix A).These commodities are selected since they are well covered by most GGCMs.In our analysis, they are also used as proxies for estimating the effect of climate change on the rest of the agricultural commodities, taking into account their plant growth characteristics (e.g.differentiating C3 and C4 crops4 ).The scenarios consider the 'middle of the road' Shared Socio-economic Pathway 2 (SSP2) and the 'high emission baseline scenario' Representative Concentration Pathway of 8.5 watts per square metre in 2100 (RCP8.5).It is important to acknowledge that the influence of CO 2 on crop yields, such as wheat, is highly influenced by management practices and climatic conditions.This relationship is more uncertain in many non-EU regions worldwide.Due to the greater uncertainties surrounding the impact of elevated CO 2 in these regions, as accounted for in the climate-biophysical ISI-MIP global simulations, we have not included CO 2 fertilization effects in the analysis.Although uncertainties in the general effects of elevated CO 2 on crops have been narrowed (Toreti et al 2020), benefits of elevated CO 2, have still to be validated, especially as interactive and yield-limiting factors such as nutrients, water, management, and extreme events, may offset benefits of elevated the CO 2 on yields (see, e.g.Helman and Bonfil 2022).
For non-EU regions, country specific yield projections from the climate-biophysical model ensembles were aggregated where necessary using the FAO regional statistics for cultivated crop areas.For the EU, the country-specific yield projections were applied at the NUTS25 regional level assuming the same country specific climate change effect.These regional crop-specific yields also take into account the limitation of water availability, separating irrigated and rain-fed agricultural production.
Baseline projections for agricultural markets for 2050 were coming from a combination of the medium-term outlook published by the European Commission, which assumes a continuation of the current EU policy framework to 2030 and global agricultural market projections from the GLOBIOM6 model to 2050.A set of baseline assumptions compliant with SSP2 were considered (Riahi et al 2017).The specific SSP2 assumptions mainly relate to GDP developments, population growth and technical change, which affect respectively food consumption, per capita income growth and yield developments.The latter in context of our analysis was already provided by the GLOBIOM projections based on the SSP2 pathway until 2050.All these SSP2 drivers are incorporated to the analysis irrespective of the climate forcing considered (RCP8.5 in this case).Consumption decisions are endogenous to the model and depend on consumer prices and demand elasticities.Additional changes in consumption (dietary shifts) as a result of climate change are not considered because the relationship between consumer behaviour and climate change is complex and not well understood (Thøgersen 2021).Moreover, the aim of the paper is to analyse the market feedback due to climate change merely from a supply side perspective.
Given the large number of scenarios, results are presented as box-whisker plots.In addition, the results from NUTS2 level are aggregated at national and regional (Northern and Southern Europe) to be able to summarize the large scenario ensemble and understand the contrasting impacts across regions.The plots display the distribution, i.e. the variability and range by crop and country around the median of the 35 different scenarios, making the box-whisker plots a suitable way to display the range of climate change impacts.Note: The boxplot displays the median, two hinges and two whiskers of a continuous variable distribution.The lower and upper hinges (coloured bars) correspond to the first and third quartiles (the 25th and 75th percentiles).The upper whisker extends from the hinge to the largest value no further than 1.5 than the inter-quartile range (IQR), or distance between the first and third quartiles.The lower whisker extends from the hinge to the smallest value at most 1.5 * IQR of the hinge.Data beyond the end of the whiskers are 'outliers' points and are plotted individually as dots.

Yield impacts
Figure 1 shows both the direct (exogenous) biophysical climate change-related yield shocks for some of the major global crop producing regions and the resulting (endogenous) yields after considering economic effects and the readjustment of domestic and international agricultural markets.As pointed out earlier, the final yield changes are endogenous in the economic model, thus the simulated yields incorporate market effects and do not necessarily need to follow the same spatial pattern as the exogenous yield shocks.
In general, the market-adjusted yield changes follow the impacts from the pure biophysical climate change-related yield shocks.The negative biophysical effects from climate change are prevailing for most of the regions and crops, with significant losses for wheat producers in Argentina, Russia and Ukraine.Climate change is not going to be equally bad for all, because there are regions and crops that will experience positive yield changes, especially regions and crops in the Northern Hemisphere (grain maize and soybeans in Canada for example).However, when considering the agro-economic market feedback some regions benefit despite the negatively projected yields.For example, Argentina is projected to benefit in a global context in case of soybean production despite the negative biophysical yield effects, Europe in the case of grain maize production and India in the case of wheat production.The economic reasoning behind such altering effects is that depending on the size of the shock farmers initially will prioritize the crops that experience positive yield effects due to climate change and produce less the crops with more negative yield effects, resulting in supply changes at the global market.The higher supply will influence producer prices, which will decrease for the prioritized crops and increase for the crops that are produced less, which in turn provides incentives to produce the non-prioritized crops in more intensively with higher inputs (e.g.fertilizers, plant protection, machinery, etc.), resulting in change in the yields and area (crop mix) which is affecting the average yields per ha.However, input prices will also react as a result of the input demand changes, consequently affecting again the share of technology variant (intensive/extensive), yields and activity levels.
As pointed out, the aggregated EU may also benefit when biophysical effects are coupled with market feedback, which also holds when examining the results at more disaggregated level (Northern vs Southern Europe). Figure 2 shows that the median crop-specific yields in Northern Europe will be positively affected by climate change, which holds for both the yield changes coming from the climatebiophysical models and the market-adjusted ones from the economic model.Some regions, such as France and Germany are expected to have significant biophysical yield increases by around 25% (median effect) for rice and soybeans (see appendix B1).Conversely, cereal producers in Southern Europe will be in general more negatively affected.Barley, grain maize and wheat are the crops most affected by climate change in Southern Europe, with sizeable biophysical yield reductions (−10%, −6% and −9%, respectively).Rice and soybeans are not an exception and most of the regions in Southern Europe will be negatively affected by climate change (see appendix B1), although the results indicate positive biophysical yield impacts on these crops at the aggregated level, which is mainly driven by Portugal and Spain for rice and soybeans, respectively (see appendix B1).
Our modelling results show that the interplay (spill-over) of climate change impacts in other important production regions outside of Europe and trade adjustments lead to considerably different changes in the endogenous (i.e.trade-adjusted) yields around the median.In Northern Europe, despite biophysical yield increases due to climate change (i.e.longer growing season and precipitation changes), endogenous yields are lower after market re-adjustments.Because increased supply will push down market prices lowering the incentives to produce these crops more intensively.However, the uncertainty (i.e.spread from the quartiles) is fairly large, especially for rice and soybeans (±15% and ±10%, respectively).In Southern Europe, wheat yields, despite being heavily affected (−9% median effect) by climate change, respond positively to the global market adjustment (around 1% endogenous median yield increase).The same is noticeable for grain maize, as a −5% direct climate change median effect is converted into a small endogenous median yield increase after market adjustments (with higher spread on the positive side).In addition, the uncertainty range is much smaller in Southern Europe compared to Northern Europe.This is mainly driven by the significant positive yields in Portugal (see appendix B2).Other regions show negative yields, but the positive endogenous adjustment on the median is still noticeable despite the overall negative yield effects.

Impact on agricultural production and income
As a result of the positive yield adjustments, agricultural production indicators (i.e.area, gross/net farm income, production level) show that the marketadjusted yield developments lead to positive effects in both Northern and Southern Europe in most of the scenarios (figure 3).It has to be noted that the area expansion for the considered crops occurs at the expense of other crops such as fodder crops in Northern Europe and oilseeds, pulses as well as potatoes (other arable crops) in Southern Europe (appendix C1).However, the crop area changes in both regions mainly occur at the cost of fallow land as area with lowest relative profitability to the considered crops.The area expansion of the total utilized agricultural is marginal in both regions and it comes from land that was previously abandoned.The area expansion is not limited by factors such as water availability or quality, which is especially important for rice producers, as these factors are not considered in the current agro-economic analysis.Therefore, the increase in area should be considered an upper bound of the effects from climate change on EU agriculture.
In Northern Europe, grain maize, wheat and rice area and production increase by about 10% (median effect).Production of barley is projected to remain stable with the exception of Germany, where it is expected to decrease by around 20% (median effect) as a result of reallocating more area to grain maize and wheat production (see appendix B2).In Northern Europe, soybean area and production are increasing the most in relative terms, especially in Austria (more than 25%, see appendix B2).However, they also display large variability around the median (20%), with increases of more than 75% in some scenarios and slight decreases in others.This large variability can be explained by the fact that soybean production in the EU is concentrated in a relatively small area (0.5% of total agricultural land; EC 2019) and, therefore, relative production changes seem quite big although they are rather small in absolute terms.EU rice production is concentrated on an even smaller area than soybeans, mainly in some regions in Southern Europe (0.25% of total agricultural land).Accordingly, the large relative changes plotted for rice and soybeans have to be interpreted with care because these impacts are rather small in absolute terms.
Similar to Northern Europe, area expansion and supply in Southern Europe is projected to increase for the five crops analysed.Area expansion is noticeable and significant (median effect) in regions with positive endogenous yields, but even some regions with negative endogenous yields show area expansion (see appendix B2).This area increase is mainly due to compensate for the losses coming from the climate change-induced yield reduction and is driven by the producer price increase.Still, the overall production effect is negative or supply shows almost no change around the median due to the area changes.Variability in grain maize area is much smaller than for wheat, which is a consequence of the type of crop (C4 vs C3, respectively).For example, in the case of Italy the negative exogenous yield shocks are similar for both crops (appendix B1), but wheat area changes are much more spread around the median than for grain maize (appendix B2).This indicates that maize as C4 crop is more resilient to climate change than wheat as C3 crop, which consequently affects the area and (endogenous) yield developments.Although the relative changes for wheat are smaller than for rice and soybeans, wheat changes are much more important because they involve a much higher volume at aggregated level (around 14 million tons for wheat compared to 0.01 million tons for rice and soybean).The same holds for grain maize and barley (around 12 and 8 million tons, respectively).Thus, their increase in area and supply in Southern Europe (4% around median) is more important because of the bigger absolute amounts comprised in the higher endogenous yields.
The changes in production and producer prices lead to a higher agricultural income in the EU.Agricultural income takes into account the changes in the product margins (gross value added less cost) and in the production quantity.The income increases range between 25% and 50% in Northern Europe and 10%-30% in Southern Europe for wheat, barley, grain maize and soybean producers.For rice producers in Northern and Southern Europe, income increases by 60% and 70%, respectively.Therefore, European farming sector are projected to experience a net benefit from global climate change when considering market adjustment.In the following section we focus on the price changes and additional drivers that underpin the increase in EU agricultural income.

Impact on prices, consumption and trade
Due to climate change and reduced supply globally of the considered crops, global market and therefore EU producer prices are projected to increase in both parts of Europe, ranging between 1% and around 7% for wheat, barley, grain maize and soybeans, and 28% for rice (median effect).(figure 4).These price effects also hold at regional level (see appendix B3).Apart from the effect on prices following basic microeconomic theory, i.e. lower supply leads to increase in prices, another reason for price increase is that the endogenous yield adjustment and more intensive input use result in higher production costs, which are also reflected in increased producer prices.Figures 5 and 6 show that the increase in production intensity in the EU is mainly driven by increased exports and not domestic use.In addition, the inelastic demand in the model also contributes to the increase in producer prices.As a consequence of the large price increase for rice, the income of EU rice producers is projected to increase by around 60% (median effect) in Northern Europe, especially France and Romania (appendix B2) and more than 70% in Southern Europe (figure 3), for example Italy, Spain, Portugal and Bulgaria.However, it can be noticed that there is high variability in the farm income as a result of the high variability in production.Consumer prices7 in both regions are of similar magnitude as producer prices in terms of absolute changes, but the relative changes are much lower due to high consumer margins which are assumed constant (figure 4).But also because we did not considered any social-economic impacts on the demand due to climate change that would directly affect the demand curves (see figure 5).
Despite the marginal relative changes in consumer prices, domestic consumption of wheat and soybean in both Northern and Southern Europe decreases (median effect over the different scenarios), whereas it increases for maize (figure 5) given that wheat and soybean consumer prices are increasing marginally compared to maize, but enough to affect consumer behaviour and expenditures.Lower domestic consumption allows for a significant increase in exports, especially from Northern Europe where median exports are projected to increase by more than 50% for wheat and 25% for soybean (figure 6 and appendix B5), but still remain at rather low absolute levels for soybean.Higher domestic consumption of maize can also be explained by an increase in EU pig meat and poultry production, and hence higher feed demand (appendix C2).The ruminants herd size is affected negatively by climate change in both regions, thus the marginally higher consumer prices for wheat and soybeans (figure 6) lead to changes from less cereals and protein to a more grass-based feed mix.In addition, grain maize domestic consumption in Northern Europe displays the largest variability in regions where livestock production is more pronounced (see appendix B4).The increase in rice demand and domestic consumption in the EU is because other countries outside of the EU (such as India, China and the USA) have stronger CC-driven negative effects on rice production (appendix D).The EU will still remain a large net importer of rice, despite the high relative changes in terms of exports, especially from Italy (see appendix B5), since the absolute quantities concerned are rather small.As a result of these changes, rice consumer and producer prices increase in the scenario (figure 4), but the large price variability is reflected in the large uncertainty (variation in model results) for domestic use.
As mentioned, the projected decline in domestic use for wheat in both regions is mainly due to the increase in exports.Wheat exports relative changes have to be considered important given that the EU is already a large wheat exporter in the baseline (23.5 million tons).The increases in producer prices indicated in the previous section are also driven by the trade effect at the expense (losses) of other wheat producers, mainly in the USA, Brazil, India and Africa (see figure 1 and appendix D).Southern Europe exports of wheat are not increasing as much as in Northern Europe, which is due to the lower endogenous yield effects (figure 2).Barley production in Southern Europe, especially from Bulgaria, Croatia and Spain (appendix B5), will be mainly exportoriented due to negative climate effects in the USA, China, Russia, and Africa (appendix D).
The higher production of cereals (figure 3 and appendix B2) allows improving the net trade balance in several EU regions for some crops, in particular by reducing imports (wheat and soybeans).Wheat imports in Austria as well as soybeans in Czechia and Slovakia are declining considerably due to the positive effects of climate change on yields (appendix B5).Barley imports in Southern Europe, however, are projected to increase.Southern European regions are mainly net importers of barley (on average) with around four times more imports than exports.For example, in the case of Croatia the increase in exports is compensated by import increases to keep demand and domestic use balanced.Hence, apart from supporting adaptation to climate change, trade also helps to buffer the socio-economic impacts from climate change.In general, compared to the baseline, the EU still remains a net exporter of wheat and barley, and a net importer of grain maize, rice and soybeans.

Discussion and conclusions
Climate change clearly poses a threat to global food production in the medium to long term, and Europe will not be an exception (IPCC 2019).However, despite the overall negative effects on crop yields and global supply, some regions and agricultural producers will benefit from climate change at the expense of others.Our analysis shows that when considering market feedback in agro-economic models, the negative effects of climate change on crop yields and agricultural productivity in EU projected by biophysical models, may be partially compensated due to the readjustment of commodity markets.Due to larger negative impacts on agricultural productivity in big non-EU agricultural production and exporting regions, the EU farming sector may be benefiting given their competitive advantage in terms of lower yield losses, especially in the Northern parts of Europe.Trade feedbacks and market re-adjustments are the main reasons for a potential increase of EU wheat and maize production by 2050.Based on our analysis, derived from a series of global gridded biophysical and climate model ensembles coupled with a global agricultural market model, it can be concluded that market adjustments are important when analysing climate change impacts in the agricultural sector.Our results are in line with the findings of Nelson et al (2014), Baldos and Hertel (2015), Blanco et al (2017) and Martinez et al (2017), who emphasize the importance of considering agricultural market interactions when analysing the overall impacts of climate change on the agricultural sector.As we consider a larger scenario ensemble and consistently different combinations of climate and biophysical models within a global context, our analysis can be considered more robust than previous ones.
Our results imply that careful management of agricultural producer practices under climate change conditions may limit yield losses, especially when considering additional export opportunities for the EU farming sector.Specifically, our results indicate that under global climate change and market adjustments, EU farming sector will adapt and could experience a net benefit due to a significant increase in agricultural income in the face of climate change.However, there are regions in Northern Europe along with most Southern Europe regions, that are expected to be negatively impacted by climate change, even after market adjustments have been made.This implies that for further adaptation to climate change, farmers need to make use of targeted management measures and practices, such as drought-resistant crop varieties and practices regarding more widely use yield or index-based insurance schemes, investments in technologies that increase water use efficiency, crop diversification, soil organic matter management, and tillage systems (Beveridge et al 2018, Hansen et al 2019, Beszner et al 2022).Regarding the latter, Pittelkow et al (2015) found in a meta-analysis that no-tillage yields match or can even be higher than the conventional yields after 3-10 years, especially in dry climates and rain-fed conditions such as the most vulnerable regions in Southern Europe.In addition, Donatelli et al (2015) showed that simple adaptation techniques (changing sowing dates and the use of different varieties) in combination with different rainfall patterns and increased photosynthesis efficiency due to CO 2 concentrations, may lead to positive productivity changes, including Southern Europe.Policy makers could focus on such measures and support respective practices for climate change adaptation in the EU.
Although we did not consider any changes in consumer behaviour in the scenario ensemble, it is important to mention that reduced consumption of food products with a lower carbon footprint in production as well as lower carbon emission during use, will offset to a certain level the effect on climate change, thus the warming and yield effects.Moran et al (2020) have estimated that implementing such changes could lead to 25% reduction of the carbon footprint in Europe.However, most consumers without assistance cannot identify the behavioral changes that are worth doing to mitigate climate change (Thøgersen 2021).Hence, we believe that the estimated effects on consumption and prices are in line with such behaviour, unless the carbon footprint is reflected in prices to assist the consumers and make climate friendly choices.As a result of the marginal increase in consumer prices it can be argued that the loss in consumer surplus in the EU is much lower than the gain in producer surplus from the higher producer prices.Consequently, the EU farming sector as a whole is likely to benefit from global climate change under the assumptions taken in this paper.It should be noted that there are many other climate change impacts that tend to affect negatively economic welfare and which cannot be considered within the current modelling framework.
An important limitation of our analysis is the potential lack of water in Southern Europe and in parts of Northern Europe, which may aggravate under climate change conditions.Even though water availability is projected to be an issue for many parts of Southern Europe, the economic analysis in this study has not taken water limitations into account, which are expected to have consequences for the entire agro-economic production system, especially when considering extreme climate events (i.e. more frequent and severe droughts).However, irrigation is a positive climate adaptation mechanism which allows to minimize losses from negative climate change yield impacts (Elliott et al 2014).Considering water availability and water as a production input was not possible in this study, mainly because in the economic model it is only considered for European regions (Blanco et al 2018) and not for non-EU regions, which would lead to inconsistencies in our analysis.Therefore, these aspects should be considered in future work.In addition, it was not possible to specifically consider the differences between the biophysical yield shocks for irrigated and rain-fed crops provided by the ISIMIP database, which is why we used an aggregated yield shock based on the irrigated and rain-fed areas.
It may be also argued that the choice of the SSP2 in the baseline is not an adequate pathway to represent future social, economic and technological developments, given the more recent and current ones.However, Fricko et al (2017) showed that this pathway reflects the historical path in terms of carbon and energy intensity improvements.It also occupies a central position between the SSPs regarding the major mitigation (average annual energy intensity improvement rate, the amount of cumulative CO 2 emissions, the estimated net present value of GDP losses), and adaptation challenge dimensions (number of people globally that have no access to clean forms of energy, the share of the global population that remains illiterate, the share of the global population that has received tertiary education).
Due to a lack of data, livestock production is not considered to be directly affected by climate change in the scenarios presented.However, the livestock sector is indirectly impacted through the effects on feed availability, which are transmitted via higher production costs to dairy and meat producers.In reality, certain climate change-related temperature increases could severely threaten livestock productivity, especially if no adaptation strategies are put in place.Accordingly, the projected small increases for EU beef, poultry and pork production that partially underpin the increases in barley and grain maize production have to be taken with caution.Moreover, it is also important to stress that the increased interannual climate variability may have more important implications for farming sector and markets than the long-term average response.However, seasonal variability is omitted in our analysis due to lack of data.We also focus on the yield effects from climate change without considering the effect of elevated CO 2 fertilization.Parry et al (2004) show that crops under certain conditions tend to experience higher yields under elevated CO 2 levels.Hence, the stronger negative shocks from increased temperature in the non-EU regions may be partially offset when considering CO 2 fertilization effects, consequently also lowering the benefits that EU farming sector may experience from market feedback.However, Long et al (2006) found that although elevated CO 2 improves yields, it may be by around 50% less than what other studies found.Müller et al (2021) found that the crop models projections agree more strongly, if the process of CO 2 fertilization is ignored.Moreover, Derner et al (2003) concluded that the potential effects on yields and plant biomass from higher CO 2 concentration depend mainly on the nutrient and water levels.Helman and Bonfil (2022) also found out that yield effect of elevated CO 2 is especially not observed in the world's top wheat-producing countries due to six decades of warming and drought effects.Thus, although our results may be more pronounced than results considering CO 2 fertilization, the degree of overestimation may be rather low.In addition, potential positive effects will likely be outweighed by water availability/scarcity, more weeds and invasive plant species from warmer climate as well as extreme events (e.g.heat waves, extreme drought).
With respect to climate extreme events, current impact assessments tend to underestimate the effects of climate extremes, as these processes are not taken into account in the productivity shocks (i.e.mean estimates).The expected increase in frequency and intensity of climate extremes as well as the projected recurrent and concurrent events in key producing regions of the world (Toreti et al 2019a(Toreti et al , 2019b) ) may trigger yield and production losses, inducing higher price variability and altering global food markets (Chatzopoulos et al 2019, Toreti and Perez-Dominguez 2019).However, these aspects are difficult to quantify with the current modelling framework, which evaluates long-term changes in average conditions.Indeed, shocks induced by recurrent and concurrent large-scale extreme events may destabilise the global production system and markets and have highly non-linear and long-term effects.
Despite these limitations, our results contribute to the better understanding of climate change impacts on agriculture, especially in the EU.With respect to the two main research questions posed in this study, our results show that market feedback is important and has to be taken into account when analysing the overall climate change impacts on agricultural production at regional level.The interplay (spill-over) of climate change impacts and market adjustments can lead to considerably different changes in the endogenous (i.e.market-adjusted) yields compared to the pure climate change-induced biophysical yield shocks.The effects can range from reinforcing, mitigating to reversing the biophysical yield impacts.Moreover, although Northern Europe will generally experience positive and Southern Europe negative impacts on agriculture due to climate change, our results underline that the range of possible impacts of climate change for EU agriculture is more revealing at regional level.In this respect, our results could help fostering a better design of policy measures for the adaptation to climate change, especially as they point out the most vulnerable regions in the EU.Moreover, our study may provide a blueprint for other comprehensive agro-economic analysis of a large ensemble of scenarios from different climate and crop models, applied globally in a consistent manner and focusing on results in specific regions.

Figure 1 .
Figure 1.Regional crop yield changes: exogenous yield shocks (biophysical) and endogenous response (economic) in 2050 relative to the baseline.Note: The boxplot displays the median, two hinges and two whiskers of a continuous variable distribution.The lower and upper hinges (coloured bars) correspond to the first and third quartiles (the 25th and 75th percentiles).The upper whisker extends from the hinge to the largest value no further than 1.5 than the inter-quartile range (IQR), or distance between the first and third quartiles.The lower whisker extends from the hinge to the smallest value at most 1.5 * IQR of the hinge.Data beyond the end of the whiskers are 'outliers' points and are plotted individually as dots.

Figure 3 .
Figure 3. Changes in crop area, production and agricultural income in Northern and Southern Europe in 2050 relative to the baseline.

Figure 4 .
Figure 4. Changes in Northern and Southern Europe consumer and producer prices in 2050 relative to the baseline.

Figure 5 .
Figure 5. Changes in Northern and Southern Europe demand and domestic consumption in 2050 relative to the baseline.

Figure 6 .
Figure 6.Changes in Northern and Southern Europe exports and imports in 2050 relative to the baseline.

B2.
Regional changes in crop area, farm income and production in 2050 relative to the baseline B4.Regional changes in demand and domestic consumption in 2050 relative to the baseline B5.Regional changes in exports and imports in 2050 relative to the baseline C2.Changes in livestock numbers in Norther and Southern Europe in 2050 relative to the baseline D2.European climate change ISIMIP Fast-Track median yield changes under the RCP 8.5 and SSP2 scenarios for the selected crops in 2050 relative to the baseline

Table 1 .
Summary of features included in the models.
no-till yield more?A global meta-analysis Field Crops Res.183 156-68 Riahi K, van Vuuren D P, Kriegler E, Edmonds J, O'Neill B C and Tavoni M 2017 The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: an overview Glob.Environ.Change 42 153-68 Shrestha S, Ciaian P, Himics M and Van Doorslaer B 2013 Impacts of climate change on EU agriculture Rev. Agric.Appl.Econ.16 24-39 Thøgersen J 2021 Consumer behavior and climate change: consumers need considerable assistance Curr.Opin.Behav.Sci.42 9-14 Toreti A et al 2019a The exceptional 2018 European water seesaw calls for action on adaptation Earth's Future 7 652-63 Toreti A, Cronie O and Zampieri M 2019b Concurrent climate extremes in the key wheat producing regions of the world Sci.Rep. 9 5393 Toreti A, Deryng D, Tubiello F N, Müller C, Kimball B A, Moser G and Rosenzweig C 2020 Narrowing uncertainties in the effects of elevated CO2 on crops Nat.Food 1 775-82 Toreti A and Pérez Domínguez I 2019 Concurrent Climate Extremes and Agricultural Shocks (Science for Policy Brief, European Commission) Van Meijl H et al 2018 Comparing impacts of climate change and mitigation on global agriculture by 2050 Environ.Res.Lett.13 064021 Von Lampe M et al 2014 Why do global long term scenarios for agriculture differ?An overview of the AgMIP global economic model intercomparison Agric.Econ.45 3-20