سیاست های مدیریت ریسک کشاورزی تحت عدم قطعیت آب و هوا

Climate change is forecasted to increase the variability of weather conditions and the frequency of extreme events. Due to potential adverse impacts on crop yields it will have implications for demand of agricultural risk management instruments and farmers’ adaptation strategies. Evidence on climate change impacts on crop yield variability and estimates of production risk from farm surveys in Australia, Canada and Spain, are used to analyse the policy choice between three different types of insurance (individual, area-yield and weather index) and ex post payments. The results are found to be subject to strong uncertainties and depend on the risk profile of different farmers and locations; the paper provides several insights on how to analyse these complexities. In general, area yield performs best more often across our countries and scenarios, in particular for the baseline and marginal climate change (without increases in extreme events). However, area yield can be very expensive if farmers have limited information on how climate change affects yields (misalignment in expectations), and particularly so under extreme climate change scenarios. In these more challenging cases, ex post payments perform well to increase low incomes when the risk is systemic like in Australia; Weather index performs well to reduce the welfare costs of risks when the correlation between yields and index is increased by the extreme events. The paper also analyses the robustness of different instruments in the face of limited knowledge of the probabilities of different climate change scenarios; highlighting that this added layer of uncertainty could be overcome to provide sound policy advice under uncertainties introduced by climate change. The role of providing information to farmers on impacts of climate change emerges as a crucial result of this paper as indicated by the significantly higher budgetary expenditures occurring across all instruments when farmers’ expectations are misaligned relative to actual impacts of climate change.

Climate change affects the mean and variability of weather conditions and the frequency of extreme events, which to a great extent determines the variability of production and yields. The risk management response to these changes is part of farmers’ adaptation strategies. Of relevance to risk management, the yield of crops is limited to differing degrees by water availability and temperature depending on the agro-ecological zone. Examples of impacts from extreme events and weather variability are the significantly increased costs resulting from the increased frequency of extremely hot days that cause heat stress in crops, or by the timing and amount of rainfall in a specific event. Climate change affects the distribution of yields under a given set of management practices, which in turn affects the probability distribution for farmers’ expected income. Farmers can adopt several adaptation strategies in response to these changes. Adaptation through cropping pattern change can in some cases ease the exposure of plants to critical higher temperatures (Peltonen-Sainio et al., 2011). Also, changing planting time may help avoid heat stress during the critical growth phases (Rötter et al., 2011). Another means is to introduce more diverse cultivars that differ genetically in their responsiveness to climate conditions (Howden et al., 2007). As regards precipitation changes and water shortage, farmers can adjust by improving soil water-holding capacity by adding crop residues or manure, or by adopting conservation tillage (Smith and Olesen, 2010 and Känkänen et al., 2011). Altering fertiliser rates to maintain grain or fruit quality consistent with the climate is another option. In a situation where farmers have no insurance, there should be in principle a strong incentive to adapt to climate change (Mendelsohn, 2010). Farmer reactions are more nuanced, however, and lack of insurance has shown that there is a lower likelihood of farmers adopting new technologies (Feder et al., 1985 and Antle and Crissman, 1990), of lower investments (Skees et al., 1999), but also of greater diversification (Skees et al., 1999). Finally, even though certain practices may decrease risk once they are mastered by the farmer, the risk of crop-failure can increase initially because changing practices can be risky as farmers learn new technologies (Marra et al., 2003). Risk management instruments, such as crop insurance and disaster assistance programme, and especially how they are designed, will affect incentives to adapt (Collier et al., 2009). For example, traditional agricultural insurance (which makes an indemnity payment when the farm incurs a verifiable production loss) can help to manage production risk but it is known to be expensive and will diminish incentives to adapt to climate change. Weather index insurance or area yield insurance, which do not require on-farm verification, can help keep administrative costs down as compared to individual yield insurance, and they do not discourage adaptation since indemnities are paid independently of actual loss incurred by a policyholder. However, they are not a means for structural adaptation. Farmers will incorporate any insurance subsidies or ex post disaster payments to their production decisions, which may favour insurance over crop diversification or other risk management and adaptation strategies. Insurance is sometimes used as a disaster assistance tool. It has the advantage of a formal contract with the financial participation of farmers, the evaluation of damages and a relatively quick payment of indemnities. But support to insurance has also its drawbacks; in particular it can prevent the development of other fully private solutions and it typically does not fully replace ex post assistance. Is climate change making insurance and other risk management policies more needed? How can policy makers take such decisions when the information about how different instruments would perform under an uncertain climate is very limited? Building on previous work examining risk management under climate change (Collier et al., 2009 and Heltberg et al., 2009) this paper is the first to address, in an applied context, the risk and the uncertainties introduced by climate change in the probability of weather events, and the role of perceptions of this uncertainty in terms of how risk management policies would perform in practice. To investigate these issues we provide examples from Australia, Canada, and Spain, which highlight that the appropriateness of a policy's design depends on how climate change affects the risk structure facing farmers. The paper also analyses the robustness of policy instrument relative to current uncertainty on the impact of climate change on variability of yields. The multidimensional, diverse and uncertain nature of the problem of risk management under climate change makes it difficult to identify an optimal policy choice. First, there is strong uncertainty about the quantitative impact of climate change on the variability of yields and production risks. Second there is uncertainty about farmer's perceived risks and their degree and direction of adaptation response to climate change. Third there is a strong farm-specific or idiosyncratic component because different farms have different risk profiles, are affected differently by climate change and have different adaptation responses. Finally the range of policy options is very large. In this paper we try to tackle each of these dimensions, respectively: analysing three climate scenarios (one standard or “marginal” climate change scenario, one with higher frequency of extreme events, and a baseline with no climate change); looking at three different responses by farmers (adaptation by diversification, structural adaptation and misalignment); characterising three types of farms according to their risk profile; and finally comparing four different policy options. This is a highly complex decision making framework where bio-physical impacts of climate change interact with the human response. The strategy followed in this paper is to use science and economic empirical analysis to try to provide insight on these interactions and their policy relevance in three countries with different characteristics. Of course the reality is even more complex with many more possible scenarios and types of farms and with very limited information about their likelihood and frequency. An alternative strategy for our research would have been to simplify the problem by reducing its dimension. However, our analysis is already a simplification and our purpose is also using this example to illustrate the difficulties of analysing highly uncertain policy questions such as those related with climate change. One important conclusion of the analysis in this paper is that scientists and economists need to address the added uncertainty introduced by climate change if they are to give sound policy advice. The results of the analysis help to understand the dimensions and trade-offs of the policy question, and possible ways to get more robust policies. The paper is structured as follows. Section 2 presents crop insurance instruments and ex post payments analysed in this paper. In Section 3, the data and the stochastic simulation model used for analysis are presented. Results under a climate change scenario without more extreme events and misalignment (“marginal” climate change scenario) are discussed in Section 4 while alternative climate change scenarios and robust policy choices under strong uncertainties are, respectively, presented in Sections 5 and 6. Section 7 concludes.

Few insights into the impact of climate change on the variability of crop yields are provided in the available literature, although there is relatively more empirical information of its impacts on the level of average yields. The impact of climate change differs depending on the location. For example, the most reliable sources to date reveal that climate change will increase production risk as measured by yield variability of the main crops in continental Spain, but that yield variability on the Canadian Prairies will likely be reduced for crops such as wheat and barley. In Australia, the evidence varies with some commodities showing increased production risk and others showing reduced risk. In the case of climate change scenarios that only marginally change the risk environment, the demand for insurance increases only marginally (except in Spain) and the crowding out of adaptation by diversification remains. Individual yield insurance tends to be very costly for governments, while weather index insurance and ex post payments are cheaper on average. Ex post payments are highly variable and can be extremely high in some years. On the whole, however, if climate change only slightly modifies yield variability then the new risks associated with climate change do not seem to be an appropriate justification or basis on which to develop new risk management policies. The analysis in this study goes beyond a standard climate change scenario and investigates policy making under strong uncertainty. First, two different climate change scenarios are examined: standard climate change versus a situation with numerous extreme events. Second, three different behavioural responses by farmers are examined: no response due to limited information on how climate change affects yields (misalignment); adaptation by diversification; and structural adaptation. The strong uncertainties about the climate change scenarios and behavioural responses (referred to as “ambiguities”) are organised in seven scenarios. Estimating the cost-effectiveness of each measure in each scenario is a complex quantitative exercise and the results are not always intuitive and differ across countries and farm types. These results highlight the usefulness of scenarios to deal with strong uncertainties. In policy problems subject to high uncertainty and diversity such as risk management in agriculture under climate change, there is not such a thing as an “optimal policy”. In order to contribute to these areas of decision making that are becoming more relevant over time, scientists have to try to find ways to analyse this complexity without taking the shortcut of oversimplifying the problem. This paper provides new insights on how to deal with these uncertainties and identify robust policies, which will vary depending on the country context and the information available to farmers of the impact of climate change. Area yield performs best more often across the countries and scenarios presented here, in particular for the baseline and marginal climate change. But Area yield can be very expensive under extremes scenarios. In this case: ex post payments perform well to increase low incomes when the risk is systemic like in Australia; Weather index performs well to reduce the welfare costs of risks when the correlation between yields and index is increased by the extreme events. The possibility of extreme events and misalignment scenarios significantly changes the policy decision environment. The analysis of government's best response to this ambiguity is very challenging and requires a significant change in the approach. Rather than identifying optimal policies, the definition and understanding of the plausible scenarios is a core part of the analysis. Governments may seek the implementation of “robust” policies that are not optimal under any scenario but that may be able to respond well to different environments and avoid very bad outcomes, particularly under extreme events and misalignment. The misalignment scenarios are characterised by high budgetary expenditure and low adaptation practices. Other policy initiatives that focus on information and training can help prevent the misalignment of risk perceptions. This study shows that it is technically feasible to define plausible scenarios and implement robust criteria in response to strong climate change uncertainties. An additional policy challenge to prevent misalignment is how to communicate complex information and uncertainties about climate change and policy implications to policy makers, farmers, insurers and other stakeholders. But this is beyond the scope of this paper.