Novel crop insurance products for rural Africa
Farm communities worldwide face difficulties when trying to insure income against price and weather fluctuations. In principle, they can choose between self-insurance, mutual insurance and market insurance.
Self-insurance and mutual insurance
Self-insurance generally operates through measures in the real sphere such as the development of irrigation and spreading of insecticides, whereas mutual insurance shares the risk among farm households and comprises both physical activities, such as helping out at harvest time or in case of an accident, and income transfers in cash and in kind. Combined, both types provide the main risk coping mechanism for farmers throughout Sub-Sahara Africa. However, these traditional instruments prove to offer insufficient coverage against aggregate risks such as those originating from droughts, floods and price shocks that threaten all members of the local communities at the same time.
Market insurance of crops
The alternative of market insurance of crops would seem a natural candidate instrument. Insurance arrangements increasingly figure on the policy making and research agenda, due to the increased attention to social safety nets as part of an overall assistance strategy, the scope to provide income transfers via insurance schemes that are compatible with WTO legislation and the introduction of risk management tools as part of agricultural policy packages in developed countries. Yet, market insurance is also plagued by excessive monitoring costs in avoiding the two classical problems of adverse selection and moral hazard (overrepresentation in the pool of insured of people with higher risks and overstating of damages, negligence of insured property, and defaulting on premium payments by the insured and on compensation payments by the insurers). In the Sub-Sahara African context it is also unrealistic to expect farmers to pay the premium ahead of the harvest. Despite these difficulties, developing countries are looking for ways to enhance market insurance arrangements in some mixed public-private form. In urban areas, targeted instruments such as health insurance and free provision of drugs have been introduced for this purpose. However, in rural areas, where few public facilities are available, more general arrangements such as crop insurance, possibly subsidized, seem preferable.        
Subsidized market insurance
As compared to purely publicly funded schemes, possibly subsidized market insurance offers the advantage that the insured groups pay some premium themselves. Besides alleviating the pressures on the already overburdened national budgets, this reduces the scope for free riding on arrangements, since premium payers will tend to exercise some countervailing pressure. Furthermore, by creating pools of policy holders with different risk profiles, say, of farmers with different cropping patterns, it offers possibilities for mutual insurance at above village level, and also for solidarity between population groups within the limits set by political realities, of course. Finally, at the international level, market insurance enjoys increased popularity because of some further advantages: since the economies in rural Sub-Sahara Africa are small by international standards, reinsurance on the international financial markets is relatively easy; also, even when subsidized, crop insurance arrangements are considered less distortive than many other farm support measures and so far enjoy a Green Box status at the WTO.
Present situation       
Yet, at present crop insurance in Sub-Sahara Africa is still practically non-existent. Consequently, poor farmers often find themselves trapped in poverty, unable to take advantage of upcoming profitable opportunities. The risk is too high for them to take a chance. This situation has become a major concern of the development community, particularly at a time that arrangements to control the markets directly, through prices and stock management, have been discarded as less effective. The search for new financial arrangements to help building social safety nets in rural Africa has led to several initiatives.
New initiatives: index-based insurance
Notably, in recent years, a new product has been piloted that seek to address problems of insurance by conditioning the indemnity payments on a set of agreed upon conditions that is independent from both farmers’ and insurers’ decisions. Such index-based insurance pays out when an agreed upon indicator falls below an agreed upon threshold. Classical examples are payments triggered by the recorded rainfall at a particular weather station or, by the price at a particular market, or by weather-price conditions simultaneously.  Ongoing research and experience indicate that it is not easy to piece together an index function that predicts well the actual individual damages on the basis of selected weather and price variables, i.e. an index function that has a low basis risk. Poor farmers tend to be reluctant to buy index-based insurance, despite significant subsidies often offered on the premiums. One reason may be that based on the indemnifications paid so far, they consider the basis risk relatively high under the proposed schemes. Another reason may be that they do not see how the current participation by their neighbors could ever overcome the problems of covariate risk they so often faced under mutual insurance arrangements.         
Improving the design
This has motivated research at SOW-VU to improve the design of index-based insurance products. In particular, a new technique has been developed to estimate an index-based indemnification scheme that minimizes basis risk – specified as the shortfall of income below a specified poverty line – , that can be specified for any pool of farmers willing to share risk, and that adapts to a given level of self-financing. The technique adapts semi-parametric techniques from Support Vector Regression, considers a set of weather and price variables, and constructs a function that is flexible and converges to the scheme that is closest to a given ideal scheme, i.e. a scheme that eliminates all basis risk and brings all farmers at or above the poverty line.        
Rural Ghana
The technique was applied to rural Ghana, where it appears that over the period 1980-2005 about half of the farm households had a living standard below the poverty line. Calculations on the historical data indicate that, with a small mark-up for the cost of monitoring and reinsurance, an ideal national mutual insurance scheme with a per hectare premium of 12 per cent of mean income could have avoided all shortfalls below the poverty line. As group-specific schemes by region or by quintiles lack the solidarity implicit in such a national arrangement, the ideal requires important subsidies from government or donors in case of a regional scheme, say for poor groups in Northern Ghana.         
Box 1: Estimated income distribution of Ghanaian farm households
The graph shows the income distribution of Ghanaian farm households estimated and simulated for 100 different types of farmers under 26 various states of the world reflecting price and weather conditions from 1980 to 2005. The fat vertical line represents the poverty line, and income shortfalls are displayed to the right of this line. The dashed line represents income shortfalls in the uninsured (historical) case, the continuous line depicts an ideal insurance to which all farmers subscribe, while the dashed-dotted line refers to outcomes under a semi-parametric index-based insurance scheme. The comparison of the incidence and depth of poverty under the various arrangements illustrates how insurance can reduce poverty. In the uninsured case the probability of shortfall below the poverty line is high (the area to the right of the poverty line is 47% of the total), while under the ideal national insurance, as expected, this probability is reduced to zero. Due to the premium payments, many farmers are moving towards the poverty line in many states of the world, and, for example, the share with an income exceeding twice the poverty line is only 13 percent, half of the percentage in the uninsured case. In the case of an index-based insurance, the shortfalls to the poverty line diminish significantly (from 47% to 28%) but obviously much less than in the ideal case. The depth of poverty is reduced as well, as can be seen from the narrowing of the right-hand side tail of the distribution.
Simulated effects of index-based insurance
Simulations with index-based insurance products that try to fit the ideal indemnity payments on rainfall and prices show, as expected, that lower rainfall and lower prices imply higher payments. Specifically, among these variables, rainfall is the best predictor, followed by the price of staple crops, which might reflect the relatively high vulnerability of staple crop farmers in the Northern part as compared to the better-off situation of cash-crop farmers in the Southern part. The simulated effects on poverty appear to be substantial, as explained in Box 1. Poverty prevalence reduced on average while also the depth of poverty decreased. It further appears that commonly employed parametric forms have a relatively poor capacity to fit the ideal indemnification scheme, whereas the corresponding semi-parametric forms perform much better.         
Individual needs of the farmers and future circumstances
Although index-based insurance seems to be a promising adaptation of classical crop insurance, an important caveat applies. By construction, index-based insurance involves basis risk giving up the option of full adaptation to individual needs of the poor. One reason is that the policy uniformly applies to all policy holders with many individuals having the same observed prices and weather, about the same farm size but, say, different cropping patterns, yields, and personal situations. Whatever its shape, an index-based insurance will only compute a single indemnity payment for all of them. The second reason is that the insurance contract has to apply in the future, under circumstances other than those found in the historical record. Therefore, it is important to avoid overfitting whereby the contract may nicely fit the past but performs poorly under conditions beyond the historical record.

This project is partially supported by the World Bank and has been executed in cooperation with researchers from the Ghana Statistical Service and the Vrije Universiteit.