Collier argues that index insurance is best suited to cover disasters which occur every 20 years or greater because severe risks lower premium costs. Below are excerpts form the interview.

Q: You have done substantial research, notably with Jerry Skees on index insurance. Why do you think that index insurance needs to be designed as “disaster insurance” rather than “crop insurance”?

A: We need a new frame of reference for household markets. Up to now, the consensus focus has been crop yields. Yet the effects of bad weather on household well-being are often multifaceted and poorly captured by yields for a single crop. What we’ve seen is households treat the insurance as an investment; if it doesn’t pay within a year or two, they stop buying. Rather than changing contracts to pay more frequently, as some have tried, we need to make contracts more relevant.

Instead, consider a frame of reference akin to life insurance — index-based disaster insurance, coverage designed and marketed for a severe natural disaster that we hope households will never experience. While the specific product design would be contextual, the focus here is events probably on the order of a 20 year return period or greater. I am sensitive to the plights of households that experience climate-related wealth shocks every three or five years, but risk reduction, not insurance, needs to be the priority for them. Households employ complex risk management strategies that work fairly well for year-to-year fluctuations, but occasionally a disaster occurs that is so bad it undermines household capacity. These events affect households similarly to the death of a breadwinner, undercutting opportunities to improve long-term household well-being through education, asset accumulation, etc. Rather than seeing it as a bad investment, life insurance policyholders typically recognize that not having to file a claim is a reason to celebrate. Moreover, focusing exclusively on severe risks lowers the cost. In sum, if index insurance were designed as disaster insurance, rather than crop insurance, I think it would be in many cases a much more effective product for households.

Q: Does this mean that this insurance product would not take into account the agronomics of specific crops?

A: The core idea is that if the production risks of a specific crop are highly related to household wealth shocks, then, yes those phenological considerations would be important to product design. Of course, the point of the response is that this situation – households having practically all their investments in a single, specific crop that demonstrates high returns except for rare climate events – is not descriptive of the vast majority of vulnerable households. More generally, agricultural production will tend to be designed around the typical weather conditions in the region and severe deviations from the norm will hurt many crops (and potentially non-farm livelihoods).

Q: You have argued that in its role of "promoting access to credit", index insurance is more effective and efficient at the meso-level (rural banks and microfinance institutions) than at the micro-level (households). Why is that?

A: Natural disasters can trigger large loan losses that create capital adequacy and/or liquidity problems for lenders. As a consequence, lenders usually avoid households and firms in vulnerable areas. This credit risk can be managed potentially with index insurance, either micro products for households or meso products for lenders. Meso products have two advantages. First, they benefit from the lender's portfolio aggregation, reducing basis risk. Second, they exploit lender information. The core business of lenders is assessing the repayment capacity of borrowers. Still, while only a portion of vulnerable borrowers tend to default even for severe disasters, it is often difficult to tell beforehand who will default. After an event, lenders use their expertise to estimate who can repay and adjust loan terms as needed. These actions minimize losses and so reduce the amount of insurance needed for lenders using meso products. In contrast, micro products do not benefit from ex post lender information and so often require all vulnerable borrowers to insure. Since only some of those borrowers would default, this approach over-insures the credit risk, increasing the cost of borrowing and so limiting credit access.

Q: But in this case, it is the bank that is protected against adverse weather-events, not the farmers?

A: Meso-insurance does not address household vulnerability directly, though it is likely to improve access to credit after a disaster and give lenders greater capacity to adjust loan terms. Still, micro products to protect credit may not address household vulnerability either, especially when they are designed to match a loan contract for a specific crop and the lender is the first claimant. Certain conditions may discourage using meso products, for instance when lender information is very poor. However, the arguments here suggest that generally, micro products to protect credit should be the concessionary result of market limitations rather than the status quo.

Q: Where does index-insurance fit in a larger public policy objective of strengthening credit markets vulnerable to disaster risk?

A: The prudential standards in most countries rely on international guidelines originally developed for large, global banks. Those standards tend to be insensitive to the risks of inclusive finance providers, failing to encourage portfolio diversification and depending on capital reserves for addressing severe events. Reliance on capital reserves motivates lenders to manage losses by contracting credit after a severe event, during a time when communities need credit to recover. Policymakers and regulators interested in increasing the resilience of the banking system to disaster risk can 1) employ rules allowing for more risk sensitive supervision of lender portfolios and 2) formally recognize a broader set of mechanisms such as diversification and financial risk transfer (e.g., index insurance, derivatives, contingent credit) for managing disasters).