Many agricultural development programs are ideally positioned to support the widespread adoption of “sustainable intensification” technologies and management practices. While the science underlying these programs may be well-established, their actual success hinges critically on their ability to overcome a myriad of institutional and implementation challenges, particularly since many operate in environments fraught with risks, notably increasingly erratic weather. This is especially true in South Asia, which has been the victim of a progressively unpredictable and variable monsoon.
Economic theory makes several unambiguous predictions regarding smallholder farmers’ decision making under risk. Without adequate strategies for managing risks, farmers tend to behave in an overly conservative fashion, storing assets in highly liquid—though lower-return—forms, and opting for lower-risk, lower-return input-use strategies. While such strategies may reduce production or profit variability, they do so via costly mechanisms that inhibit a household’s long-term growth trajectory. An obvious implication of the theory is that mitigating risks should lead to a reversal of these predictions, encouraging greater investments in higher risk, higher returning activities.
One method for mitigating these risks is insurance, which allows farmers to transfer a portion of their risk to the insurer. While theoretically workable, attempts to provide formal, indemnity-based crop insurance have struggled for various reasons. Most notably, conventional insurance suffers from high administrative and transaction costs, as well as asymmetric information between insured and insurer, which in turn gives rise to adverse selection and moral hazard.
Consequently, many policy makers and practitioners have turned to index-based insurance programs, which base their payments on the performance of some transparent, easy-to-measure index relative to some benchmark—such as weather data or average area yields. But because insurance payments are based on index performance rather than direct on-field assessments, index insurance results in a nontrivial probability that farmers will not be compensated even when they experience significant losses. This residual risk—known as basis risk—has been identified as a significant impediment in many index insurance programs piloted around the world.
A recent IFPRI Discussion Paper presents results from a randomized controlled trial designed to assess both the demand for and effectiveness of an innovative index insurance product in rural Bangladesh. The insurance product was designed to help smallholder farmers manage risks related to droughts during the aman rice-growing season (monsoon season). While many drought index insurance products have been piloted in the past, the current product was designed to address multiple dimensions of drought risk, enabling farmers to manage risks associated with both diminished crop yields as well as increased production costs. To that end, the insurance design incorporates a weather index to offset risks associated with increased production costs—particularly utilization of supplemental irrigation—due to prolonged dry spells, with an accompanying area yield index to protect against risks associated with yield losses, regardless of the cause of the losses.
Demand for the index insurance product was found to be quite price-sensitive, and consistent with empirical evidence from both developed and developing countries around the world, we observed very little insurance take-up without incentives (in the form of discounts and rebates). But these incentives were largely successful in increasing insurance coverage, with discounts resulting in higher coverage in aggregate relative to rebates.
Despite theoretical and empirical evidence that present bias and other factors may constrain demand, we find that much of the variation in take-up rates that can be explained is due to the type and level of incentives, with other characteristics and preferences explaining relatively little variation in demand. One notable exception is risk aversion. We found evidence that more risk-averse individuals purchase lower index insurance coverage, likely due to their aversion to residual basis risk. Offering these individuals a rebate instead of a discount could lead to increased insurance coverage, as they implicitly viewed rebates as a commitment savings mechanism that can offset the costs associated with basis risk, especially if they experience on-farm losses yet are not indemnified by the insurance.
Perhaps more interestingly, we found evidence of both ex ante risk mitigation effects as well as ex post income or liquidity effects. Ex ante, risk mitigation increased agricultural production along the intensive margin during the aman season. In particular, insurance resulted in increased agricultural input expenditures for aman rice production, with the total increase exceeding the average potential insurance payouts. Notably, however, the increase in expenditures was not distributed uniformly over all inputs, but reflected investments biased in favor of several modern agricultural inputs, notably irrigation, fertilizers, and pesticides. In percentage terms, the increases in input expenditures are striking. Insured farmers invested 23 percent more on fertilizers than they would have without insurance, which, other things being equal, increased farmers’ yield potential. Insurance coverage on the cost of production risk also resulted in a roughly 25 percent increase in investments in irrigation, likely indicative of efforts to mitigate the yield impact of prolonged dry spells that were recorded in the 2013 aman season. We also observed a 22 percent increase in pesticide use, suggesting that even with drought risks ameliorated by the index insurance product, the problems of moral hazard that plague traditional indemnity-based insurance programs do not arise.
Ex post, the additional income realized from the insurance payouts prior to boro (dry season) land preparation led to increased production along the extensive margin. In particular, the disbursement of insurance payments provided farmers with a liquidity injection that led to an expansion of boro area and increased expenditures on seeds, resulting in an overall increase in boro rice production. While we were not able to ascertain precisely whether the increase in seed expenditures was solely due to the increase in area or whether it reflected farmers investing in new seeds as part of a quest for higher yields, such investments nevertheless increase farmers’ exposure to technological improvements embodied in newer varieties. While this ex post income effect likely only arose due to the insurance payout (since there was essentially no effect on aman yields or total production) we posit that a similar income effect could be realized even under situations in which the insurance payouts are not triggered. Such an event would occur under more optimal weather conditions, and the increased investments in modern inputs—in particular fertilizers and pesticides—arising from the ex ante risk mitigation effects should result in higher yields, which in turn could generate a similar ex post income effect. Since we have limited exogenous variation in insurance performance to exploit, this remains merely a plausible hypothesis.
The results from this study have important policy implications for promoting agricultural risk management as a means to a more productive and profitable end for agricultural development in Bangladesh. The study provides convincing evidence that index insurance can provide farmers with the peace of mind needed to encourage investments in higher risk, higher return agricultural production. While there may be need for government interventions to subsidize the cost of insurance, the long-term benefits of these public expenditures may exceed their costs.
Patrick Ward is a Research Fellow in IFPRI’s Environment and Production Technology Division.
