【正文】
damage risks using the DB CV method. We find around half of the sample households was willing to buy an insurance in principle which indeed indicates low demand for crop insurance. Majority of those respondents who did not agree to buy an insurance indicated lack of money ine as reason for not buying insurance. This finding confirms that ine constraint is a major demand side obstacle to setup a disaster insurance market in developing countries. However, a considerably large proportion (33%) of the respondents who did not want to buy disaster insurance in principle indicated that they did not like the terms and conditions of the proposed insurance scheme. This finding is important to take on board while designing a microinsurance scheme for rural population in Bangladesh as this finding, to some extent, reflects that a pletely conventional type of insurance market may not be popular among the target clients. Our study reveals that crop insurance demand varies across household head’s primary occupation, land ownership and size of the farm land. We find that it was mainly agricultural farmers who owned large parcels of farm land were willing to buy crop insurance to protect themselves against the risks of catastrophic damage. Our study further reveals that crop damage cost and household WTP to reduce crop damage vary significantly across the nature of the disaster risks. Households who suffered from the highest average crop damage costs per catastrophic event were willing to pay the lowest premium to avoid crop damage risks. We show that such paradoxical result in estimated WTP arises due to the existing disparity in average household ine in different risk areas. However, it is important to note that the demand analysis presented in this paper is primarily based on observed associations and relationships using linear correlations and nonparametric testing procedures. A more extended deterministic model to further test the underlying causal relationships and their directions was beyond the scope of the current study. On the basis of the crop damage cost data obtained from the household survey, we calculate expected average indemnity payment by the potential insurance providers on the basis of two different indemnity payout principles, namely forgone revenue ine and production cost. Using average crop damage cost incurred by farm households in different risk areas, we tested our simple analytical model of mercial viability by paring the future value of the expected premium receivable by the insurer with the expected indemnity payable to the insured. Assuming zero administrative cost and 10% interest rate per annum, we find crop insurance schemes are marginally viable in riverine flood plain areas. The discrepancy between average expected indemnity payment and future value of expected insurance premium is too large to be converged by the variation in damage cost estimation or the indemnity payout function. Three important policy implications follow from the results presented in this study. First, the findings of the study suggest that a uniform structure of crop insurance market does not exist in Bangladesh. Crop damage varies depending on the nature of catastrophe risks and WTP varies depending on the socioeconomic characteristics of rural munities living in different risk areas. Hence, crop insurance scheme needs to be developed carefully by taking these two key criterions into consideration. Second, the feasibility test results presented in this paper demonstrate that the crop damage risks faced by the farm households living in the wetland basin and the coastal floodplain are not insurable. The estimated indemnity payable by the insurers, in these two risk areas, consistently exceeds the expected insurance premium receivable. The estimated discrepancies between expected indemnity and expected premium seem too large to be financed by government subsidy on a co