Designing Crop Insurance to Help Farmers Transfer Risk of Crop Loss in Rural Indonesia

Transcription

1 Designing Crop Insurance to Help Farmers Transfer Risk of Crop Loss in Rural Indonesia by Brigitta Aryanti A Second Year Policy Analysis written in fulfillment of the requirements for the degree of Master of Public Administration in International Development, John F. Kennedy School of Government, Harvard University Client: Ministry of Agriculture, Republic of Indonesia Advisor: Brigitte Madrian Seminar Leader: Rema Hanna March 17, 2014

2 Acknowledgements I would like to thank my advisor, Brigitte Madrian and my SYPA seminar leader, Rema Hanna, for their valuable advice, feedback, and encouragement at all stages of the SYPA. I would also like to thank my MPA/ID family for their continuous support throughout the program. I got by with a little help from my friends. I would especially like to thank my classmates in SYPA Section B for the stimulating discussions and valuable inputs given for both my SYPA writing and presentations. Finally, I would like to thank my husband, Aryo, and my daughter, Kirana, for their patience, support, and understanding during the entire MPA/ID experience. Your smile and love are the reasons I survive! I should note that Indonesia s Ministry of Agriculture s interest and/or involvement in this paper are only hypothetical. The findings, interpretations, and conclusions expressed in this report are those of the author and should not be attributed in any manner to Harvard Kennedy School. All errors remain my own. 1

3 Table of Contents Acknowledgements... 1 Table of Contents... 2 List of Figures... 3 List of Tables... 3 Executive Summary... 4 A. Introduction... 5 B. Agriculture in Indonesia: The Context... 8 C. Risk Assessment for Smallholder Farmers in Indonesia... 9 a. Risks Faced by Rural Households, Causes, and Consequences b. Household Existing Risk Coping Mechanisms D. Characteristics of Households Most Vulnerable to Crop Loss E. Determining The Suitable Crop Insurance Model For Indonesia a. Current Crop Insurance Model: Multiple-peril Crop Insurance (MPCI) b. Alternative Insurance Model I: Damage-based/Single Peril Crop Insurance c. Alternative Insurance Model II: Area Yield Index Insurance d. Alternative Insurance Model III: Crop Weather Index Insurance (WII) e. Suitable Crop Insurance Option for Indonesia F. Political Economy of Crop Insurance in Indonesia a. Ministry of Agriculture (MoA) b. Financial Services Authority (Otoritas Jasa Keuangan/OJK) c. Legislature d. Insurers e. Farmers G. Recommended Actions H. Conclusions References Annexes Annex 1. IFLS 1-4 Data Explanation Annex 2. Regression Results Annex 3. Features of Crop Insurance Products

4 List of Figures Figure 1. Regional Distribution of Farm Households in Indonesia.. 9 Figure 2. Risks Reported by Rural Households Figure 3. Percentage of Households by Cause of Crop Loss (2007) Figure 8. Suitability and Feasibility Assessment of Crop Insurance Models for Indonesia Figure 9. Mapping of Stakeholders in Indonesian Crop Insurance.. 25 List of Tables Table 4. Average Ratio of Cost to Farm Income Per Household by Farm Size (2007) Table 5. Average Ratio of Financial Loss to Farm Income Per Household by Type of Crop (2007) Table 6. Major Crop Loss Coping Mechanisms for Rural Households.. 14 Table 7. Potential Crop Insurance Products for Indonesia.. 23 Table 10. Recommended Actions for Crop Insurance Development Keywords: crop insurance, Indonesia, area-yield index insurance, weather-index insurance, risk transfer, IFLS 3

5 Executive Summary Approximately 50.2 percent of the Indonesian population resides in rural areas, where agriculture is the main source of income for 63.8 percent of them. These households are exposed to agricultural risks that cause adverse impacts on their income and wellbeing, which makes them stuck under poverty; a condition often referred as poverty trap. However, despite the importance of agriculture, as well as its associated risks, Indonesia lacks a national agricultural insurance scheme. This paper examines the risks, causes, consequences, and existing coping strategies of farm households in rural Indonesia to support the Ministry of Agriculture to determine the need for crop insurance as well as outline the appropriate design. Using household level data from the Indonesian Family Life Survey, the paper finds that the risk most often reported by farm households is crop loss. The main cause of crop loss is drought and pestilence, although the causes of crop loss differ by island. In terms of economic loss, drought and pestilence cause the most damage. However, it is important to note that economic losses differ across households with different crops. The most common coping mechanism for crop loss at the household level is taking an extra job. Other important responses are reducing household expenses, taking loans and selling assets. Households that are most vulnerable to crop loss have the following characteristics: (i) have rice or corn as their main crops, (ii) are located outside of Kalimantan, (iii) have experienced a crop loss in the past, and (iv) have not experienced a natural disaster in the last five years. These characteristics differ if we breakdown households by the main crop cultivated. Since different crops are associated with different losses and different household characteristics, it is reasonable to recommend different insurance schemes for different crops. Our assessment finds that area-yield index insurance is suitable and feasible for rice farmers in Indonesia while weather-index insurance is suitable and feasible for corn farmers. Still, much preparation is needed to ensure their readiness for implementation. 4

6 A. Introduction Approximately 50.2 percent of the Indonesian population resides in rural areas where agriculture is the main source of income for 63.8 percent of them. 1 Poverty is increasingly concentrated in these areas, with 16.6 percent of rural people classified as poor compared with 9.9 percent of urban populations. 2 Households engaged in agricultural production face a number of risks, for example crop yield risks due to variance in the level or timing of rainfall or due to natural disasters, animal mortality due to infectious livestock diseases, and changing output prices. The effects of these risks may be two-fold. First, they can push some households into a cycle of poverty, especially households that are just above the poverty line. Second, risk-averse individuals may make investment decisions that reduce risk exposure but also reduce potential for income gains and wealth accumulation. Thus, agricultural risk contributes to the poverty trap experienced by rural people in developing countries. The risk most often reported by rural and farm households in Indonesia is crop loss. 3 This is not surprising considering the country s location in one of the most active disaster hot spots. Recent data reported by the Directorate General of Food Crops indicate that total production loss due to flood, drought, and pests and diseases was 2.33 million tons, or 3.9 percent of the total production in This production loss will translate into income loss for agricultural households. In this context, crop insurance would be a strategic policy response since access to formal risk transfer instruments, such as insurance, can help farmers transfer excessive losses to a third party, thus stabilizing household income, and ultimately improving their welfare. However, despite the importance of agriculture in the economy 1 Badan Pusat Statistik website, 2 Badan Pusat Statistik website 3 Gabriella Berloffa & Modena, F. (2009), Income Shocks, Coping Strategies, and Consumption Smoothing: An Application to Indonesian Data, Dipartimento Di Economia, Discussion Paper No , Trento: Universita Degli Studi Di Trento, p Directorate General of Crop Protection, Estimated Data on Area Affected by Flood, Drought and Pest and Disease Infestation, Ministry of Agriculture, Jakarta, (various data sheets, ) from Sahat M. Pasaribu (2010), Developing rice farm insurance in Indonesia, Agriculture and Agricultural Science Procedia 1, p

7 and its exposure to risks, no agriculture insurance is offered in the market because of market and regulatory impediments, such as the presence of systemic risk, informational asymmetries, low risk awareness, insurance culture, and affordability among farmers, lack of infrastructure support and enabling regulatory framework. 5 These impediments justify government intervention in the provision of crop insurance. In Indonesia, the government s efforts to establish an agricultural insurance scheme have failed although an Agricultural Insurance Task Force has been established three times in the past: in 1982, 1984, 1985 respectively. 6 Several constraints to the development of agricultural insurance were encountered, including a lack of awareness on the part of farmers, the general inability of farmers to pay premiums, and lack of agricultural insurance regulations. 7 The political will to establish such a scheme re-emerged in 2008 with the Indonesian Center for Agriculture, Social, Economic and Policy Studies (ICASEPS), a unit under the Ministry of Agriculture, conducting an initial study on Agricultural Insurance with assistance from the Food and Agriculture Organization (FAO). 8 As a follow up to this study, ICASEPS conducted a pilot project in 3 districts in 2012: Tuban and Gresik, both located in the East Java Province, and Ogan Komering Ulu Timur in the South Sumatera Province. 9 These initial pilots were deemed quite successful and a second-phase-pilot is being planned to better understand the implementation challenges as well as the demand side take up in other parts of the country. The objective of this study is to assist ICASEPS in the process of designing a national agricultural insurance scheme, in particular to determine the most suitable crop insurance scheme for the Indonesian context. This study is timely given the recent passage of Law 19/2013 on Farmer Protection and Empowerment in August 2013 that 5 Olivier Mahul & Stutley, C. (2010), Government Support to Agriculture: Challenges and Options for Developing Countries, Washington DC: The World Bank. 6 ICASEPS (2009), Peluang Pengembangan Asuransi Pertanian di Indonesia, Warta Penelitian dan Pengembangan Pertanian Vol. 31 No , p International Finance Corporation, Weather Index Insurance for Maize Production in Eastern Indonesia: A Feasibility Study, Jakarta: IFC, pg Kementerian Pertanian, Asuransi Usaha Tani, Alternatif Lindungi Petani. 9 Bambang Sayaka & Pasaribu, S. M., Risk Management in Rice Farming in Indonesia: a Pilot Project for Agricultural Insurance. 6

8 mandates the government, both central and local, to protect farmers from the losses of crop loss by providing agricultural insurance. The agricultural risk management literature suggests that the first step in developing an appropriate risk management framework is to gain a clear understanding of a specific risk and how it affects agricultural livelihoods in a particular region. Since the same risk can have different effects in different contexts, a risk assessment gathers information about the specific risk and its effects and uses this information to estimate the social costs and benefits of implementing different policy alternatives for addressing a specific risk in that context. The insights of a particular risk assessment is the starting point in considering means to facilitate the development of efficient and effective risk management. Based on this framework, this paper addresses four key questions: 1. What are the main sources of crop loss and what are the income impacts for farmers? 2. How do farmers manage the risk of crop loss? 3. Who among the farmers are the most vulnerable to crop loss? 4. What is the most technically sound, administratively and politically feasible crop insurance scheme that the Indonesian government should implement to help farmer households transfer the risk of crop loss? This study finds that area-yield index insurance is suitable and feasible for rice farmers in Indonesia while weather-index insurance is suitable and feasible for corn farmers. We further recommend four action plans to be completed as a follow up to the study findings: (1) Feasibility study for area-yield index insurance, (2) Establishment of a Crop Insurance Technical Support Unit to provide specialized services to agricultural insurance companies and other risk-pooling vehicles, (3) Farmers & farmer groups education on crop insurance, and (4) Improvements in database systems for weather and crop yield. This paper proceeds as follows. The next section describes the context of agriculture in Indonesia. The third section presents a description of the risks faced by rural households, their causes and consequences, as well as existing coping 7

9 mechanisms. The fourth section describes the characteristics of households most vulnerable to crop loss. The fifth section describes the current crop insurance design that is being developed by the Ministry of Agriculture and explores the other alternative designs available, such as single peril crop insurance, area-yield index insurance, and weather index insurance (WII). The sixth section analyzes the political economy of crop insurance in Indonesia. The seventh section recommends implementation actions as a follow up to the findings. Finally, the last section concludes. B. Agriculture in Indonesia: The Context Indonesia is the world s largest archipelago with 17,000 islands that span two biogeographic regions the Indomalayan and Australasian. Indonesia s tropical climate, geography, large forested areas, and extensive coastline support the world s second highest level of biodiversity after Brazil. 10 Indonesia experiences dry climatic conditions and droughts during the warm phase of the El Nino Southern Oscillation (ENSO) cycle with significant consequences for rice output, rural incomes, and staple food prices. 11 In addition, Indonesian production is highly dependent upon rainfall. Only 17 percent of the country s cultivated area has access to irrigation infrastructure, and only 10 percent of this land is effectively irrigated while more than 80 percent of the agricultural activity depends on rainfall for irrigation. 12 The agricultural sector in Indonesia comprises of large plantations (both stateowned and private) and household production modes. The large plantations tend to focus on commodities, which are important export products (palm oil and rubber), while the farmer households mostly produce rice, corn, soybeans, fruits and vegetables Rosamond L. Naylor & Mastandrea, M. D. (2009), Coping with Climate Risks in Indonesian Rice Agriculture: A Policy Perspective, p Rosamond L. Naylor et al (2007), Assessing risk of climate variability and climate change for Indonesian rice agriculture, Stanford University: Centre for Environmental Science and Policy, May International Finance Corporation, p. v 13 Badan Pusat Statistik, Sensus Pertanian

10 Figure 1. Regional Distribution of Farm Households in Indonesia Number of agricultural households Source: Sensus Pertanian 2013, BPS Based on the 2013 Indonesian Agricultural Census (Sensus Pertanian 2013), there are 26.1 million farm households in Indonesia mainly concentrated in Java, parts of Sumatera, and in South Sulawesi (Figure 1). Of these farm households, 87.6% are smallholder farmers (owning less than 2 hectares of land) and 55.5% are subsistence farmers (owning less than 0.5 hectares of land). The majority of farm households own between hectares of land. The subsistence farmers are concentrated in the same provinces as farm households, although they are also highly concentrated in Bali, Nusa Tenggara, and Papua. C. Risk Assessment for Smallholder Farmers in Indonesia One particular step before designing a solution to manage risk is to understand the origin and nature of the risk being considered. As explained in Hardaker (1997), 14 there is a need for information on risk, its cause, its characteristics (distribution, frequency and correlation with other risks), and its consequences on farm income and on the capacity of various strategies to reduce income risk. 14 J.B. Hardaker et al (2000), Coping with Risk in Agriculture, CAB International in OECD, Income Risk Management in Agriculture, p

11 This part of the paper will try to understand the origin and nature of the risks experienced by smallholder farmers, as well as existing coping mechanism at the household level. For this purpose, we will use data from the Indonesian Family Life Survey (IFLS) because it contains the self-reported incidence of shocks, including crop loss, and responses to these shocks. The Indonesian Family Life Survey (IFLS) is an on-going longitudinal survey in Indonesia that has been conducted in 4 waves: IFLS1 in 1993, IFLS2 in 1997, IFLS3 in 2000 and IFLS4 in The sample is representative of about 83 percent of the Indonesian population and contains over 30,000 individuals living in 13 of the 27 provinces in the country. This study focuses mainly on Book II of the survey, which focuses on the Household Economy, particularly the household economic shock section in IFLS1, IFLS2, and IFLS3 and the farm business section in IFLS 15 (further explanation of each survey is provided in Annex 1). a. Risks Faced by Rural Households, Causes, and Consequences The first three rounds of IFLS showed that on average 54.1% of the rural households in the sample experienced at least one economic shock in the past 5 years. The risk most often reported in all three rounds is crop loss, followed by sickness, death, and price falls (Figure 2). On average, 18.4% of rural households reported having experienced a crop loss in the last 5 years, while 12.4% reported sickness of a household member, 10% reported death of a household member, and 8.5% reported a fall in household income due to falling crop prices. In the IFLS4 survey, 14.7% of the total rural household sample experienced a crop loss While IFLS1, IFLS2, and IFLS3 had rather similar sets of questions focusing on household economic shocks in general (type of shock, when it happened (year and month), the costs of overcoming the shock), IFLS4 focuses only on crop loss. 16 The results are quite similar with the previous waves although the time frame asked in the question was different (12 months as opposed to 5 years). 10

12 Figure 2. Risks Reported by Rural Households IFLS3! IFLS2! IFLS1! Crop loss! Sickness! Death! Price falls! Unemployment! Disaster! 0%! 10%! 20%! 30%! 40%! 50%! 60%! 70%! Source: Author calculation from IFLS1 (1993), IFLS2 (1997), IFLS3 (2000) Note: This graph shows the percentage of rural households reporting risks Crop loss itself is caused by different factors. Figure 3 shows the main causes of crop loss based on data from IFLS4. 17 The main cause of crop loss in Indonesia is drought/lack of water and pestilence/rodents. On average, 50.3% of the rural households experiencing crop loss reported drought to be the cause while 32.8% said pestilence/rodents is the cause. The causes of crop loss differ by island. While most of the reported drought, flood, and pest losses came from respondents in Java, drought was also common among respondents in Bali, flood among respondents in Sulawesi, and both pest and disease among respondents in Sumatera. Figure 3. Percentage of Households by Cause of Crop Loss (2007) Other, 8.9%! Flood, 6.2%! Disease, 10.1%! Drought/ lack of water, 49.8%! Pestilence/ Rodents, 29.0%! Source: Author calculation from IFLS4 Note: This graph shows the percentage of rural households experiencing crop loss 17 IFLS1 and IFLS3 are not used in this part of the analysis because the question was not asked in the survey. Data from IFLS2 show a similar pattern to IFLS4 and therefore is not presented. 11

13 Crop loss may have an impact on other dimensions of households welfare, for example the study by Cameron & Worswick (2001) found that in the face of a crop loss; Indonesian families with girls are more likely to reduce educational expenditure than families with boys. However, this kind of long-term impact information was not captured by the survey as the survey only captures the short-term material consequences of the crop loss. Table 4 shows the average ratio of cost to income, obtained by dividing the cost occurred to the household experiencing crop loss with the annual revenue the household received from the farm business (including production for own consumption or giving to others). In general, the cost of crop loss on average is highest if caused by drought and pestilence. The cost of crop loss due to drought is on average 37.9% of a household s annual revenue from farm business. It should be noted, however, that there is variety across households as measured by the standard deviation. One possible source of variation is the different farm sizes, however, as Table 5 shows the variation across different farm sizes is not quite evident. Table 4. Average Ratio of Cost to Farm Income Per Household by Farm Size (2007) All <2 ha >2 ha average std. dev average std. dev average std. dev drought flood pest disease other Source: Author calculation from IFLS4 Another hypothesis could be that the cost variation comes from the different crops (Table 5). Crop loss due to droughts causes the highest cost across the main crops: rice, corn, and soybean, although the average impact may be higher for corn farmers. Drought causes corn farmers lose up to 70.7% of farm revenue on average 12

14 while rice and soybean farmers lose 43.2% and 56.5% of farm revenue respectively. In addition, crop loss caused by pestilence is costly for rice farmers. Table 5. Average Ratio of Financial Loss to Farm Income Per Household by Type of Crop (2007) rice corn soybean average std. dev average std. dev average std. dev drought flood pest disease other Source: Author calculation from IFLS4 b. Household Existing Risk Coping Mechanisms In the absence of formal insurance, households have developed a variety of mechanisms to smooth consumption in the midst of uncertainty, either ex ante mechanisms, ex post mechanisms, or a combination of both. Ex ante mechanisms consist of savings, crop diversification, and plot diversification. Ex-post measures include borrowing, selling assets, or using informal insurance. Table 6 shows the percentage of rural households that use different measures in response to crop losses for the period 1993 and % of rural households report taking an extra job to overcome a crop loss in The number decreased to 29.0% in 2000 probably because more coping strategies were added in the survey, particularly strategies that relate to starting/expanding a business. Other important responses are reducing household expenses, taking loans and selling assets. Many of the coping strategies have limitations. Taking extra jobs and cutting down expenses are a logical consequence of income loss. However, if the extra jobs involve taking adolescent children out of school as labor to augment family income, then there will be long-term welfare effects. Similarly, reduced expenditures on some nonessential items such as clothing and social functions will not have much welfare effect. 13

15 However, reduced expenditures on essential items such as food, medical treatment, or child education, will have adverse short- and long-term consequences. Informal insurance is of limited value when the shock affects most people in the area, since all households will be in need of assistance at the same time. Taking loans also has its limitations. Loans from family may be less available in the face of common shocks while loans from moneylenders come at very high cost. This condition emphasizes the need of crop insurance among the impacted households. Table 6. Major Crop Loss Coping Mechanisms for Rural Households Crop loss coping strategies Number of households (1993) Percent Source: Author calculation from IFLS1 (1993) and IFLS3 (2000) Note: Because of multiple responses, percentages sum to more than 100% Number of households (2000) Percent Extra job % % Loan % % Sell assets % % Saving % % Family assistance % % Reduce expenses % % Start, expand, or change N/A % business Pray N/A % Do nothing N/A % D. Characteristics of Households Most Vulnerable to Crop Loss This section investigates the association between household characteristics and location and the probability of reporting a crop loss. This paper adapts, with some modifications, the logistic regression model used by Tesliuc & Lindert (2004) to investigate the association between households characteristics and location and the probability of reporting a shock in Guatemala. 18 The dependent variable is the 18 Tesliuc, E. & K. Lindert, Risk and Vulnerability in Guatemala: A Quantitative and Qualitative Assessment, Social Protection Discussion Paper 408, World Bank,

16 probability of a household reporting crop loss, where the value is 1 if the respondent experienced a crop loss in the last 12 months before the 2007 survey and 0 if it did not. The predictors of the model are the economic status of the household (captured by annual revenue from farm business, household size, literacy of the household head, TV ownership, access to electricity, and type of crop), exogenous characteristics of the household head (his or her age and gender), the location of the dwelling (area of residence urban or rural and region Sumatera, Java, Kalimantan), and whether the household experienced crop loss in the previous survey (IFLS3-2003) or a natural disaster in the last 5 years. The results of the logistic regression are provided in Annex 2 Part A (1). The results show that variables related to economic status of the household (household size, literacy of household head, TV ownership, and access to electricity) have a weak association with the probability of reporting a crop loss, except for type of crops. Households with rice or corn as their main crops are more likely than those with other main crops to experience a crop loss while households that cultivate soybeans are less likely to experience a crop loss. The region variables show that households located in Kalimantan are less likely to report a crop loss. Finally, households that reported a crop loss in the past are more likely than those who did not to experience a crop loss in 2007 while households that experienced a natural disaster in the last 5 years are less likely than those who did not to experience a crop loss, holding other variables constant. One possible explanation might be that there is a low presence of household farmers in disaster prone areas. These characteristics differ if we breakdown by the main crop cultivated (Annex 2 Part B). For rice farmers, past experience of crop loss are associated with a higher probability of crop loss while being located in Kalimantan and having experienced a natural disaster in the last 5 years are associated with a lower probability of crop loss. For corn farmers, past experience of crop loss is also associated with a higher probability of crop loss while experience of natural disasters is associated with a lower probability of crop loss. Finally, for soybean farmers, past experience of natural disaster and being located in Java and Bali are associated with a lower probability of crop loss. 15

17 Linear probability model is used to understand the magnitude of the association between the different variables and the probability of reporting a crop loss (Annex 2 Part A (2)). The regression result shows that households who experienced a crop loss in the previous survey are associated with a 6.5% higher probability of reporting a crop loss compared to households who did not, holding other variables constant. Meanwhile, households who experienced a natural disaster in the last five years are associated with a 6.9% lower probability of reporting a crop loss compared to households who did not, holding other variables constant. In terms of the type of crop cultivated, rice farming is associated with a 7.2% higher probability of reporting a crop loss compared to non-rice farming while corn farming is associated with a 4.9% higher probability of reporting a crop loss compared to non-corn farming, holding other variables constant. On the contrary, soybean farming is associated with a 9.6% lower probability of reporting a crop loss compared to nonsoybean farming. Finally, region-wise, households in Kalimantan are associated with a 13.1% lower probability of reporting a crop loss compared to households living outside Kalimantan. E. Determining The Suitable Crop Insurance Model For Indonesia This section provides an overview on the current crop insurance model developed by the Ministry of Agriculture and contrasts it with the alternative insurance models available in international agricultural insurance markets. Annex 3 provides a summary of the main features, benefits, and challenges of these crop insurance products. a. Current Crop Insurance Model: Multiple-peril Crop Insurance (MPCI) The crop insurance currently being developed by Ministry of Agriculture, Asuransi Usaha Tani Padi (AUTP/rice farm insurance) is a multiple-peril crop insurance that focuses on rice farms. Multiple-peril crop insurance (MPCI) is insurance where an 16

18 insured yield (typically between 50-70% of the average yield on the farm) is established, as a percentage of the historical average yield of the insured farmer. For AUTP the damage threshold is 75 percent. 19 MPCI is suitable where the damage to crops is complex, because many perils interact (for example rainfall and disease) making it difficult to determine the exact cause of loss. 20 MPCI is also suitable for drought, which develops gradually over a crop season. 21 Since AUTP covers flood, drought, and pestilence, the choice of MPCI for Indonesia seems reasonable. In MPCI, if the realized yield is less than the insured yield, an indemnity is paid equal to the difference between the actual yield and the insured yield, multiplied by a pre-agreed value of sum insured 22 per unit of yield. The settlement of the sum insured in MPCI can be done in two forms: (a) based on the valuation of the guaranteed yield at the future market price for the insured crop at the time of harvest; or (b) based on an agreed value. In the face of a crop loss, the insurance company designates loss adjusters to make a yield assessment for the determination of losses prior to the harvest. For AUTP, the sum insured is determined based on the average cost of production per hectare, which is Rp 6,000,000 or USD 500 per hectare. 23 The insurance premium of AUTP is 3 percent of the sum insured, which is Rp 180,000 or USD 15 per hectare, which for administrative reasons is fixed and common for the entire country. 24 Because most of the farmers cannot afford to pay the premium, the government will subsidize 80% of the premium, hence the government pays USD 12 per hectare and farmers pay USD 3 per hectare. 25 In the AUTP design, the insurance is given to a group of farmers (kelompok tani/poktan) rather than to individual farmers and covers the land collectively 19 Kementerian Pertanian (2013), Rencana Implementasi Asuransi Usaha Tani Padi (AUTP) Tahun 2014, a presentation for Dit. Financial Services and SOEs Bappenas on 25 October 2013, p World Bank (2007), China: Innovations in Agricultural Insurance Promoting Access to Agricultural Insurance for Small Farmers, May 2007, p World Bank (2007), p A sum insured is the amount specified in the policy up to which the insurer will pay indemnities should the insured peril(s) occur and result in a loss to the insured property (Source: FAO, 23 Kementerian Pertanian (2013), p Pasaribu, p Kementerian Pertanian (2013), p

19 cultivated by the members of the group. The rationale behind this grouping of farmers in homogeneous risk areas is to make insurance feasible for small farmer communities and reduce costs. 26 Although MPCI is widely practiced world wide, the international experience has often been poor, mainly due to low uptake, high adverse selection and moral hazard, high administrative costs, underwriting results which have generally been negative, and systemic losses in severe drought or flood years. 27 Adverse selection is the tendency of high-risk farmers to buy and maintain insurance, while others with relatively lower risk do not. Moral hazard is the risk that farmers act carelessly or irresponsibly in their farming activities because they are protected from the consequences. To avoid adverse selection, information about actual yield history and farming practices at the insured unit level is required, while to avoid moral hazard issues, individual field pre-inspections and close monitoring during the crop season is required. 28 AUTP avoid these risks by requiring farmers to practice good agricultural practices and enforcing it by penalizing farmers who do not meet these terms and conditions and fail to harvest. 29 The activities to avoid the asymmetric information problems translate into high administrative cost because a huge staff of technicians is required to perform these activities at the individual farm level and the expertise needs to be imported from overseas given there is no field level pre-inspection expertise in the country. These factors make this model rather unattractive for developing countries that are dominated by smallholder farms and lack farm-level yield data. This high administrative cost also leads this product to be highly dependent on premium subsidies from the government, another barrier for developing countries with limited resources. The Ministry of Agriculture has implemented a small pilot program in Tuban and Gresik, East Java, and Ogan Komering Ulu Timur, South Sumatra to test the AUTP model covering 3,000 hectares of rice farm. The program operates through an umbrella 26 World Bank (2011), Weather Index Insurance for Agriculture: Guidance for Development Practitioners, Agricultural and Rural Development Discussion Paper 50, November 2011, p World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, July 2009, p World Bank (2010), Guyana: Agricultural Insurance Component Prefeasibility Study Report, May 2010, p Pasaribu (2010), p

20 agreement between the Ministry of Agriculture and Jasindo (a state-owned insurance company) with the government subsidizing 80% of the premiums. 30 The pilot resulted in a loss of Rp million or USD 43,600 or a loss ratio of 467% of the premium because the pilot was not able to achieve the law of large numbers, which the initial premium design was based upon. 31 Based on an evaluation done by Ministry of Agriculture, 32 there are still many problems in the AUTP implementation: 1. There was very limited product knowledge among farmers; therefore, further socialization and training are required. 2. The quality of service was low mainly due to low quality in human resources. 3. The current design ignores local agricultural conditions and equally treats all farmers across the country. 4. The evaluation recommended an exploration of other models of insurance given that the market price of crops fluctuates, increasing the risk for insurance companies that participate in the market. As a response to the initial evaluation recommendation, the next part will explore the potential of alternative insurance models to be implemented in Indonesia, namely damage-based/single peril crop insurance, area yield index insurance, and crop weather index insurance. b. Alternative Insurance Model I: Damage-based/Single Peril Crop Insurance In damage-based indemnity insurance, the damage is measured as a percentage loss and this percentage is applied to the agreed sum insured for the crop. The sum insured is the amount specified in the insurance policy, up to which the insurer will pay indemnities, if insured peril occurs (actual yield falls short of threshold yield). The insurance premium is the monetary amount paid by the insured (rice farmer) to the insurer (crop insurance company or agency) for the relevant rice crop period and is 30 Sayaka & Pasaribu 31 Yessi Supartoyo & Kasmiati (2013), Asuransi Pertanian sebagai Alternatif Mengatasi Risiko Usaha Tani Menuju Pertanian Berkelanjutan: Tinjauan Konseptual. 32 Information obtained from written materials from Dit. Financial Services & SOEs Bappenas 19

21 usually decided as a percentage of the sum insured. In damage-based indemnity insurance, physical loss or damage to the crop is measured in the field soon after the insured damage occurs and the claim is usually settled shortly after the time of loss. Single peril crop insurance is the best-known type of indemnity-based crop insurance that has operated for many years in Europe, North America, Australia and New Zealand. 33 The key advantage is its simple design and its low operation cost, which makes it able to be offered in low rates with minimum or no government subsidy. 34 This model is most applicable for farms with single or a limited number of discrete events perils (hail, windstorm, and frost) and therefore is not suitable for the perils that affect the main crops cultivated in Indonesia. 35 c. Alternative Insurance Model II: Area Yield Index Insurance Area yield index insurance is insurance where the indemnity is based on the realized average yield of an area such as a county or district, not the actual yield of the insured party. The insured yield is established as a percentage of the average yield for the area. An indemnity is paid if the realized yield for the area is less than the insured yield regardless of the actual yield on a policyholder s farm. This type of index insurance requires historical area yield data. The key advantages of this model are that moral hazard and adverse selection are minimized, thereby reducing the administrative costs, offering the potential to market this product at lower costs to farmers. 36 The main disadvantage is that an individual farmer may incur severe losses due to localized perils i.e. hail, or flooding by a nearby river, but because these localized losses do not impact the county or departmental average yield, the grower does not receive an indemnity. 37 One of the success stories for the area-yield index insurance is the National Agricultural Insurance Scheme (NAIS) in India. The program is targeted at small and marginal farmers (with less than 2 hectares) who are highly dependent on access to 33 World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p World Bank (2011), Weather Insurance Index, p World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p

22 seasonal crop credit with crop insurance being compulsory for borrowing farmers and voluntary for non-borrowing farmers. 38 The administrative costs are kept to a minimum by linking insurance with rural finance. As a mainly compulsory program, the NAIS scheme is the world s largest crop insurance program, currently insuring about 20 million Indian farmers (representing an insurance uptake rate of about 18 percent of all farmers). 39 The main requirement for an area-yield index program is to have an objective and accurate system of measuring average crop yields for each selected crop within the defined geographical area or insured unit. 40 Additionally there should be a minimum of years of historical yield data on which to calculate: (a) the normal average yield for the selected crop; (b) the insured yield coverage level, expressed as a percentage of the average yield; and (c) the underlying pure loss cost rates associated with each coverage level. 41 It is important to note that index insurance is a new concept and their long-term sustainability as not yet been proved. d. Alternative Insurance Model III: Crop Weather Index Insurance (WII) In weather index insurance (WII), the indemnity is based on realizations of a specific weather parameter measured over a pre-specified period of time at a particular weather station. The insurance can be structured to protect against index realizations that are either so high or so low that they are expected to cause crop losses. An indemnity is paid whenever the realized value of the index exceeds or is less than a prespecified threshold. WII is common in covering drought and excessive rainfalls. The development of weather index insurance products requires meeting some preconditions in terms of weather data quality to construct the index, ensure accurate rating and risk transfer, and minimize basis risk. 42 This data requirement includes: (i) historic weather data covering years which includes extreme risks; (ii) uninterrupted daily data for index design and rating purposes; (iii) data integrity; (iv) 38 World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p World Bank (2009), Feasibility Study for Agricultural Insurance in Nepal, p World Bank (2010), Guyana Agricultural Insurance, p

23 availability of a nearby weather station for a buddy check or for a backup; (v) reliable settlement mechanism; (vi) integrity of the recording procedure; and (vii) little potential for measurement tampering. 43 Based on a feasibility study by IFC (2012) for WII on maize production in Eastern Indonesia, it is feasible to develop weather index insurance for Indonesia. 44 The study also found that WII may offer a promising approach to insure maize production in Indonesia, but the instrument is likely to apply only to specific crops in specific areas. The model is feasible to be implemented in Indonesia because there is a high quality of weather data and the length of the time series of most weather stations and a strong support from the weather agency. The agency responsible for collecting and storing weather data in Indonesia is the Badan Meteorologi Klimatologi dan Geofisika (BMKG). BMKG manages 120 manned meteorological stations, 83 automatic weather stations (AWS), 55 automatic rain gauges (ARG) and 4,500 manual rain gauges (MRG) across Indonesian territory. Other institutions, such as the Ministry of Agriculture, the Ministry of Public Works, local government offices and other research centers also manage weather stations, mostly manual rain gauges. e. Suitable Crop Insurance Option for Indonesia Table 7 provides a summary of the advantages and disadvantages of each type of crop insurance. Based on the advantages and disadvantages, the suitability and feasibility of each crop insurance model was assessed. A suitability score was assigned between 1-3 (1 being the lowest and 3 the highest) to each crop insurance model based on the type of damage covered, type of insured crop, and design suitability for smallholder farmers. Similarly, a feasibility score was assigned based on the administrative cost, readiness of infrastructure (including data availability), and readiness of human resources. 43 World Bank (2010), Guyana Agricultural Insurance, p International Finance Corporation, p

24 Table 7. Potential Crop Insurance Products for Indonesia Type of Crop Insurance Product 1. Multiple Peril Crop Insurance (MPCI) 2. Single/Named Peril Insurance 3. Area-Yield Index Insurance 4. Crop Weather Index Insurance Basis of Insurance and Indemnity Advantages % Damage Suitable for complex perils (rainfall interacting with disease), drought % Damage Normally does not require government subsidy Simple design Area-Yield Index Weather Index Low risk of moral hazard and adverse selection Low administrative cost Suitable for smallholder farmers (India case) Suitable to cover drought and excessive rainfall Suitable for specific crops in specific areas in Indonesia (i.e. maize in Eastern Indonesia). High quality of weather data available in Indonesia Disadvantages High risk of moral hazard and adverse selection High administrative cost, requires government subsidy Loss due to fluctuation of crop prices Low product knowledge Low quality of human resources Type of perils not suitable with the Indonesian context (hail, windstorm, and frost) Basis risk Require objective and accurate system of measuring average crop yields for each selected crop Requires 10 to 15 years of historical yield data Basis risk Insurers need capacity building Figure 8 provides the result of the assessment. The assessment shows that single peril insurance and MPCI are strictly dominated by both area-yield index and weather-index insurance. Given the variety of crops across Indonesia, it is reasonable to have different insurance schemes for different crops. Therefore, we recommend that area-yield index insurance be developed for rice farmers while weather-index insurance developed for corn farmers. However, many preparations needs to be done to ensure its readiness to be implemented, for example for area-yield index insurance, an objective 23

25 and accurate system of measuring average crop yields must be prepared and, for both products, farmer education is extremely important given the issue of basis risk. Figure 8. Suitability and Feasibility Assessment of Crop Insurance Models for 3! Indonesia FEASIBILITY! 2! 1! Single peril! WII! MPCI! Area-yield index! 0! 0! 1! SUITABILITY! 2! 3! F. Political Economy of Crop Insurance in Indonesia This section provides a political economy analysis of the stakeholders, as it is one of the main determinants of the success or failure of crop insurance programs. There are 4 main stakeholders in crop insurance: regulators/controllers, providers/implementers, facilitators/influencers, and beneficiaries. This paper will focus the stakeholder analysis on several key actors: Ministry of Agriculture and Financial Services Authority as regulators, insurers both state-owned and private as providers/implementers, and individual farmers and farmer groups as beneficiaries. Figure 9 shows the positions of each actors based on their interests and influence. The three actors on the upper-right quadrant (Legislature, Financial Service Authority, and Ministry of Agriculture) are key players that hold strategic position in the success of crop insurance. Therefore, a strong and good relationship must be built with all of them. The three actors on the lower-right quadrant (Providers, Local Governments, Farmers) may be source of risks and therefore require careful monitoring and management. 24

26 Figure 9. Mapping of Stakeholders in Indonesian Crop Insurance Note: Red shows risk of decline in interest a. Ministry of Agriculture (MoA) The Ministry of Agriculture (MoA) has shown strong interest in crop insurance, by actively advocating and preparing Law 19/2013 on farmer empowerment, and initiating crop insurance pilot projects and studies. MoA will also be responsible to publish the general guide for crop insurance. This guide will be used by local government at provincial and district (kabupaten) level to issue technical guidance of the implementation in the field. 45 Pasaribu (2010) explained the general policy framework for rice farm insurance in his paper as follows. This policy framework should be relevant to any kind of crop insurance implemented. The first step in implementation of a national rice farm insurance policy is establishment of National Rice Insurance Commission (NRIC) for internalizing, approving and announcing the policy. It should be a high-powered national authority and Minister for Agriculture should be its chairman. In this connection, the Ministry of Agriculture is suggested to take initiative by establishing the National Rice Insurance Commission (NRIC). The Minister s decree may be needed to establish such a commission. The NRIC should consist of several competent individuals drawn from different institutions and related organizations, including the Ministry of Finance, National Development Planning Agency (Bappenas), and Ministry of Home Affairs. The 45 Pasaribu (2010), p

27 NRIC will propose a final policy package to the Ministry of Agriculture for decision. The maximum sum insured, premium rates, and premium subsidy will be determined by the NRIC. The policy package will also include the name of a national level agency or organization, which will implement the scheme on behalf of the Indonesian Government. This may be named as Agricultural Insurance Agency of Indonesia (AIAI). b. Financial Services Authority (Otoritas Jasa Keuangan/OJK) The IFC (2012) feasibility study for WII stated that the Insurance Bureau of OJK was supportive for the development of index insurance in Indonesia. The main interest of the organization is product compliance, financial system stability, and consumer protection. As the sole regulator of the insurance industry, OJK is mandated to authorize the retail distribution of all insurance products, and in order to do that, the Insurance Bureau needs to be provided with clear evidence that the product complies with the regulations, particularly the requirements for marketing a new insurance product as specified in Article 3 of MoF Decree No. 422/KMK.06/ In that regard, the Insurance Bureau particularly stated that the elaboration of the contract structure should be carried out by professionals with a recognized expertise in the field and should be based on scientifically sound methodologies. 47 In addition, the pricing process should also be carried out according to standard actuarial practices and should result in the determination of an adequate premium level to avoid placing the financial stability of the insurance company at risk and loading the policy holders with excessive burdens. 48 The premium price should not be discriminatory, which means that two covered parties within the same area of coverage and same risk exposure should be charged same rates. 49 The Insurance Bureau also requires that policyholders be educated about the functioning of the policies International Finance Corporation (2012), p International Finance Corporation (2012), p International Finance Corporation (2012), p International Finance Corporation (2012), p International Finance Corporation (2012), p