Managing Basis Risk with Multi-scale Index Insurance _____________________________________________________________ March 30, 2013 Abstract Well-designed agricultural index insurance contracts should aim to not only stabilize farm incomes, but to also stimulate investment and increase farm profits. When an index is not sufficiently correlated with losses, and thus basis risk is high, the potential for an index insurance scheme to increase the expected income of farmers is lower. We analyze the impact of basis risk on the demand for index insurance under risk and compound risk aversion, a concept akin to ambiguity aversion. We simulate demand for index insurance using field experiments conducted in Mali and show that basis risk further decreases demand under compound risk-aversion. We then propose an innovative double-scale index insurance contract that considerably reduces the amount of basis risk. We again use simulations to demonstrate that this double-scale contract increases insurance uptake. Finally, we report and discuss the actual uptake of this contract in the context of cotton farming in Mali. Keywords: Index Insurance, Microinsurance, Crop Insurance, Risk and Uncertainty JEL Classification Codes: D81, O12, O16, Q12, Q13 1. Introduction Exposure to risk, much like death and taxes in Benjamin Franklin’s famous quip, is one of life’s few certitudes. Even for those of us who are fortunate enough to live in developed countries, where a host of financial instruments are available to hedge against the vagaries of life, there will always remains a certain amount of risk which we are exposed to, because some risks are simply uninsurable (Gardner, 2009). Indeed, many contracts include a force majeure clause which renders a contract null and void in case of an act of war or terrorism, or in case of an act of God. This uninsured risk is commonly referred to as basis risk, i.e., the risk that remains when using imperfect hedging. For people who live in developing countries, where underdevelopment almost always extends to financial markets and where financial instruments to hedge against risk are fewer and further between, basis risk is even more prevalent. This is especially so for smallholder farmers, whose livelihoods are directly dependent on the risk and uncertainty associated with shifting agro-ecological conditions and largely unpredictable weather patterns. The presence of basis risk means that those smallholders often sink valuable resources which could otherwise be employed productively into hedging against risk (Scott, 1977), that they refuse to adopt new technologies designed to improve their productivity, or that they enter inefficient contracts that serve to as partial insurance mechanisms (Stiglitz, 1974). The rise of microfinance over the last 20 years has brought with it the development of financial instruments designed to protect the poor in developing countries against some of the basis risk they face. 1 Those financial instruments are generally referred to as “microinsurance” (Morduch, 2006). This paper looks at index insurance, i.e., microinsurance contracts wherein payouts from the insurer to the insured are tied to some measure that is easily verifiable by the insurer and cannot be tampered with by the insured – the so-called index. The insurance contract we consider in this paper was designed for Malian cotton farmers, whose cotton harvests are highly variable. The specific index this contract is tied to is the cotton yield in a given area (e.g., village, district, region, etc.). This is measured very precisely at various local and regional levels in Mali because of the industrial organization of cotton in that country, where the entire cotton production is purchased from producers by a parastatal. Equally importantly, the area for which the index is measured can be set at a level which individual cotton producers cannot tamper with. The former guarantees that everyone knows when the insurance pays out and when it does not. The latter guarantees that moral hazard is all but eliminated. In practice, the insurance pays out if the average cotton yield in the area falls below a specific threshold. 1 As Roodman (2011) notes, though microfinance is most famously associated with microloans, microfinance also has two other components, viz. microsavings and microinsurance. 2 In this paper, we describe an innovation we came up with when designing our index insurance contract which allows us to further reduce the amount of baseline risk West African cotton farmers are exposed to. Put simply, that innovation adds a second index to the insurance contract. Thus, the insurance pays out if two requirements are met: (i) the average cotton yield in a given producer’s area falls below a certain threshold, and (ii) the average cotton yield in a given producer’s cooperative falls below another threshold. We discuss the reasons why this allows eliminating basis risk in the second half of this paper. In a nutshell, however, relative to single-index insurance contracts, double-index contracts completely eliminate false positives (i.e., situations wherein a cooperative with a yield above the area threshold receives a payout) while eliminating many false negatives (i.e., situations wherein a cooperative with a yield below the area threshold does not receive a payout). This paper also describes experimental work related to index insurance which we are conducting in Mali. To do so, we first discuss a series of field experiments aimed at eliciting their degrees of risk and compound risk aversion, the latter concept being akin to ambiguity aversion (Abdellaoui et al., 2011). We then discuss the results of the first round of an ongoing randomized controlled trial (RCT) aimed at assessing the impacts of index insurance contracts on the welfare of West African cotton producers. 3 2. Risk Rationing and Cotton Production in Mali In Southern Mali, most farmers grow a mix of subsistence crops and cotton. Cotton is their main (and often their only) source of cash. Cotton is a profitable but risky crop: due to erratic rains and pests, cotton yields fluctuate a lot from year to year. Low yields translate in low farm revenue and often financial difficulties, as farmers rely heavily on credit to finance their cotton production. In a sample of 505 farming households surveyed in 2006 and 2007, all cotton growers had received an input loan from the parastatal.2 This is hardly surprising, as the input package represents a large investment compared the net revenue obtained from the activity, and so few cotton growers would be able to undertake that investment without a loan from the parastatal. 3 Loans are part of an exclusive contract between farmers and the parastatal cotton company, the Compagnie Malienne des Textiles (CMDT), which has a monopsony on cotton in Mali. The CMDT organizes the distribution of inputs and the purchase of the crop. The lender is a bank that provides loans to groups of cotton growers with a joint liability clause. The bank has an agreement with CMDT stating that a group’s cotton revenue is directly transferred to the group’s account. As a result, the joint liability clause is enforced and a group’s revenue is first used to pay back its loan. The group then has to internally solve issues related to the compensation of good 2 The data was collected by the University of Namur’s Centre de Recherche en Économie du Développement with funding from the Agence Française de Developpement. Out of the 505 households, 301 were surveyed in 2006, and 204 were surveyed in 2007. 3 In our sample for example, the average cotton revenue was 138,000 FCFA/ha while farmers bought an input package amounting to approximately 70,000 FCFA/ha, or 51% of the average cotton revenue. 4 producers who effectively paid for their co-members’ debt. If a group defaults, it faces exclusion from future loan. Our survey indicates that exclusion is not immediate, as some groups who had defaulted on their loan in the past were still granted loans, but we also surveyed two villages where CMDT and the bank had stopped their activities as a result of bad credit records. In fact, the collateral risk of loans appears to discourage many farmers from growing cotton altogether, leading to risk rationing in the credit market (Boucher, Carter and Guirkinger, 2008). Risk rationing occurs when lenders, constrained by asymmetric information, shift so much contractual risk to the borrower that the borrower voluntarily withdraws from the credit market even when she has the collateral wealth needed to qualify for a loan contract. The survey introduced above allowed obtaining an estimate of the extensive margin of risk rationing. Specifically, the questionnaire included a non-borrower module designed to elicit the reasons why they had no loans. Table 1 reports these reasons and indicates that, of all farmers who were not planting cotton (and thus not borrowing), 42% believed they could obtain a loan to grow cotton if they so wished. 4 Of these farmers who had thus access to credit but chose not to borrow, 28% declared that they were discouraged by the collateral risk of the loan. (The other leading explanation was that they did not have a profitable investment to make that required a loan.) This represents a lower bound estimate of risk-rationing, as it ignores the intensive margin: exposure 4 The remaining 58% believed they could not obtain a loan and the two main explanations were that their fields were not well suited for cotton production and that they had pending debt with their cooperative. 5 to collateral risk also implies efficiency losses for borrowers who chose sub-optimal loan sizes, and thus cultivated areas under cotton. The survey does not allow quantifying this effect, but in-depth qualitative interviews conducted with a restricted number of farmers showed that some of them recently reduced their planted area as a result of being exposed to substantial collateral risk.