Estimated damage to crop losses on French farms
(1) Dorothée Kapsambelis - Département R&D Modélisation, CCR
(2) David Moncoulon - Département R&D Modélisation, CCR
(3) Jean Cordier - Institut Agro Rennes-Angers, Economics and Management, UMR SMART
INTRODUCTION
Climatic risks cause losses in quantity and quality of crop production on farms and weaken their economy [1]. Over the 1980-2004 period, the average annual amount of damage following weatherrelated events was €1.2bn in Metropolitan France [2]. This amount is €1.95bn over the 2015-2020 period. Climate change is thought to be one of the main causes of this significant increase in the average annual loss ratio [3]. The amount of damage is all the more important during an extreme event, such as the drought of 2003, for which the damage was estimated at €4 billion for Metropolitan France [4]. This type of event is the most complex to manage for all risk stakeholders, as its systemic nature means that it does not allow for losses to be pooled at country level.
In France, in order to better support farm income in case of extreme weather events, a risk management reform is being developed. In this context, the Act of 2 March 2022 provides for a three-tier system in which each player - FarmerInsurer-State - bears part of the claim. Quantifying the necessary government allocation and capital requirements for insurers requires a costing of extreme weather events that includes the impact of climate change. This article presents the modelling chain developed by CCR: it integrates the development of a new agroclimatic index representing the hazards of drought and excess water at a fine scale on the territory, a damage model allowing the simulation of crop losses at departmental level and a downscaling allowing the evaluation of economic damage at farm level. This impact model is also used to make projections to 2050 to consider the effect of climate change.
This work was the subject of a thesis, that was defended on 12 July 2022, in partnership with the Institut Agro-Rennes and Météo-France, under the supervision ofJean Cordier and David Moncoulon.
METHODOLOGY
Data used
The data on plant production is derived from two sources:
- AGRESTE, which references yields by department and by crop from 1989 to 2020;
- the public FADN (Farm Accountancy Data Network), which presents the production developed on a representative sample of farms that have a standard gross production of at least €25,000 in Metropolitan France. This sample is made up of approximately 7,000 farms each year over the 2000-2019 period, the geographical location of which is scaled to the administrative regions.
Climate-related data on precipitation and potential evapotranspiration are taken from the SAFRAN database over the 1989-2018 period on an 8 x 8 km grid across Metropolitan France. The climate projections for 2050 are based on ARPEGE-Climat simulations with constant climate provided by Météo-France according to the IPCC RCP 8.5 emissions scenario. The study focuses on straw cereals (soft winter wheat and winter barley) and grassland.
Modelling the impact of climate change on crop losses by department
The link between climate and crop losses is achieved by creating a new climate index based on the cumulative ten-year water balance anomaly, which represents extreme drought and water loss events. This index is correlated to crop losses and is calculated on a grid of 8 km by 8 km in Metropolitan France. This grid is cross-referenced with the graphical land reference register (RGP/ référentiel parcellaire graphique) to take land cover into account. The model developed makes it possible to assign to each index value (in mm) a departmental crop loss value per crop (%) by adjusting two damage curves based on historical data, making it possible to predict the occurrence of claims and the amount of losses in the event of a claim.
The ARPEGE-Climat model from MétéoFrance is then used to model the impact of climate change. This model simulates 400 replications of the target year 2000 and 400 replications of the year 2050 under the IPCC RCP 8.5 scenario. This method allows a wide range of possible scenarios including extreme events, which are rare by definition. The climate-related variables of precipitation and evapotranspiration are output from the ARPEGE-Climat model, from which the index is recalculated, and the department and national crop losses simulated.
Downscaling
To simulate crop losses at farm level, a downscaling methodology between AGRESTE and public FADN databases is established by studying the distribution of individual crop losses versus national crop loss average. This study allows the construction of a second-order polynomial relationship between the standard deviation of individual crop losses and national average crop losses.
Once this relationship is established, crop losses at farm level are simulated using a Weibull distribution centred around the parameters of mean and standard deviation. This methodology makes it possible to analyse the distribution of individual crop losses over the 2000 and 2050 climates by crop and to deduce a risk profile by farming sector.
Figure 1A - Change in the average values of the agroclimatic index between Climate 2000 and Climate 2050 (RCP 8.5).
Figure 1B - Change in the ten values of the agroclimatic index between climate 2000 and climate 2050 (RCP 8.5)
RESULTS
The hazard - The example of drought
Figure 1A shows the average annual change in the agroclimatic index (DOWKIsech) between Climate 2000 climate and Climate 2050: for the future climate, a widespread dry-out is observed over Metropolitan France. The water deficit is even more acute around the Mediterranean and the Aquitaine Basin. Figure 1B shows the evolution of the tenyear values of the water balance anomaly. The results show that the ten-year drought increases in intensity by 2050. The cumulative water balance anomaly is tending to become more homogeneous in mainland France, with a more pronounced trend in the northern half of France (the Paris Basin), which produces a lot of cereals.
Financial damage nationwide
Figure 2 shows the economic losses of farms for the three crops studied in Climate 2000 and Climate 2050 as a result of drought and excess water events. The average annual economic losses are €1.2bn in 2000 and €1.5bn in 2050, i.e. a 22% increase. Of these damage amounts, grasslands account for about 70% of the claims. Ten-year losses increased by 34% between Climate 2000 and Climate 2050 from €2bn to €2.6bn. These results show that whatever risk management system is put in place, it must have the capacity to quickly build up reserves to compensate producers and cope with extreme events, the frequency of which is expected to increase by 2050.
# extreme events
# crop losses
# climate change
# modelling
# financial damage
Estimation of individual losses by production sector
Figure 3 presents the distribution of farmers’ crop losses over Climate 2000 and Climate 2050 for soft winter wheat over an average, 10-year, 20-year and 50-year event by class of individual crop losses. The thresholds chosen to illustrate this distribution of crop losses are 25% (current crop insurance deductible) and 50% (extreme crop loss threshold). On an average event (5% crop loss over Metropolitan France), 90% of farmers report losses of less than 25%, 50% of which have no loss at all in current and future climates. This means that these events are local and not systemic. On extreme events, the number of farms experiencing losses increases with the impact of climate change. For a 50-year event, only 11% of farms do not suffer losses. However, there are few extreme losses over 50% (less than 3% of farms). The national damage figure is therefore the sum of moderate crop losses at farm level, and these extreme climate risks intensify in terms of spatial coverage in future climate. For cereals, extreme climate risk affects a large number of farmers, with more than 70% of them suffering losses. However, these losses are moderate. Indeed, very few farmers suffer losses of more than 50%. On the scale of Metropolitan France, this type of risk is described as horizontal and systemic: it is the accumulation of moderate losses on many farmers that leads to significant economic damage. These results illustrate the difficulty of modelling risk on the scale of Metropolitan France. They also highlight the fact that extreme risk management should not focus on very high levels of individual losses but on cumulative losses.
The same analysis is carried out on grasslands and presented in Figure 4. On an average event, 30% of the farmers don’t record any losses. This figure is much lower than the one obtained for cereals, but nevertheless, the average risk remains poolable because it is not systemic. On the other hand, over the 50-year events, the number of farmers with extreme losses (>50%) changes compared to the climate in 2000 and represents more than 20% of farms. Moreover, farms without crop losses are marginal. These droughts are systemic, and their intensity is extreme. Thus, for grasslands, extreme droughts have a profile comparable to that of straw cereals in the current climate. In the future climate, however, the risk profile changes. On the scale of Metropolitan France, the amount of damage is the result of both a high level of risk (all farmers are impacted) with high individual crop losses. This risk profile is described as horizontal and vertical.
Figure 2 - Farmers’ economic losses (€M) on soft winter wheat, winter barley and grassland crops in Climate 2000 and Climate 2050 according to their return period
Figure 3 - Share of farmers (%) according to their level of crop losses over Climate 2000 and Climate 2050 for an average, 10-year, 20-year and 50-year event for soft winter wheat.
Figure 4 - Share of farmers (%) according to their level of crop losses over Climate 2000 and Climate 2050 for an average, 10-year, 20-year and 50-year event on grassland.
THE PARTNERS
The “SMART-LERECO” Joint Research Unit (JRU) was officially created on 1 January 2017, in the wake of a scientific project that brought together economists from the SMART JRU in Rennes and the UR LERECO in Nantes. This JRU brings together research and teaching in economics applied to agriculture, agri-food and the environment. The research focuses on two main areas: Organisations, Structures and Performance (OSP) and Agricultural Production and Markets.
CONCLUSION
This study models the impact of climate change in terms of economic damage on agricultural production down to individual level. This innovative methodology, targeting extreme events, allows risks to be forecasted and quantified, which is necessary to assess an appropriate and sustainable risk management system in the future. The results show an increase in exposure to extreme dryness risks for straw cereals and grasslands. Within this scope, the €2.6bn damage threshold would be exceeded every 10 years on average if all French farms were insured. The analysis of crop loss distribution at farm level reveals a different risk profile across different commodities.
For straw cereals, extreme climate risk is horizontal and systemic: the entire agricultural sector is affected, and crop losses are generally moderate. This profile continues to 2050. For grasslands, extreme climate risk is horizontal and systemic with the current climate and will evolve into both a horizontal and vertical profile in the future: all agricultural professions are affected, and individual crop losses are vast. These results highlight the difficulty of pooling these risks. To face the challenge of climate change, agriculture will have no choice but to adapt, and the new risk management system put in place will have to take this adaptation into account at individual level./
REFERENCES
1. Cordier, J., Erhel A., Pindard, A., and Frédéric Courleux. 2008. “La gestion des risques en agriculture de la théorie à la mise en oeuvre: éléments de réflexion pour l’action publique”, 40.
2. OCDE 2012. Managing Risk in Agriculture: Policy Assessment and Design OECD. https://doi. org/10.1787/9789264116 146-en.
3. Boyer, Philippe. 2008. “Assurer les calamités agricoles ?” Notes et Etudes Economiques, No. 30 (March). https://hal.archives-ouvertes. en/hal-03043063.
4. IPCC. 2022. “Climate Change 2022 : impacts, adaptation and vulnerability”. Cambridge University Press. Cambridge, UK, and New-York, NY, USA.
5. Létard, V., Flandre, H. and Lepeltier, S. 2004. “La France et les Français face à la canicule : les leçons d’une crise”. Information report to the French Senate 195. https:// www.senat.fr/rap/r03-195/r03- 19510.html
CITATION
Kapsambelis, D., et al, Damage estimates on crop losses of French farms in the context of climate change. In CCR 2022 Scientific Report; CCR, Paris, France, 2022, pp. 30-33