One of the most popular responses to drought – and disasters more generally – by aid agencies today is insurance. This fits the current development mood, requiring market-based solutions that operate at a distance and work seemingly ‘efficiently’, offsetting the need for cumbersome, late responses of state or development agency delivered humanitarian aid.
It all sounds good in theory, but what is the reality? This was a theme we explored in Zimbabwe both in the field with farmers across our sites, as well as in Harare with some of those managing a new insurance approach for drought response, sponsored by the Africa Risk Capacity Group, and supported by multiple donors.
Many types of insurance
Insurance comes in many shapes and forms. There’s classic indemnity insurance where if a disaster strikes an assessor will estimate the damage and pay out if you have a policy. This is the sort of insurance you have in case your house burns down or you crash your car.
Index-based insurance by contrast aims to pay out before the event happens. This provides early response and avoids the dangers either of later payment or of emergency responses that can undermine people’s livelihoods in the longer term. Here pay-outs occur if a threshold is crossed, say of rainfall or its proxy, such as vegetation cover. Index-based policies have become quite popular for agricultural and livestock insurance in developing countries, for instance.
A final type is sovereign insurance where a whole country takes out a policy against drought (or some other disaster) and the pay-out happens again if a threshold (the ‘attachment point’) is crossed (such as below average rainfall across the country, assumed to be affecting a certain number of people). This is aimed at ensuring ‘early action’ as part of an anticipatory approach to assistance, which again doesn’t have to wait for the mobilisation of international funds after the event.
Sovereign insurance: the Africa Risk Capacity model
This ‘sovereign insurance’ is what the government of Zimbabwe has recently bought from ARC (USD 2.5m) in partnership with the UN World Food Programme (USD 1.5m) and the START network (USD 2.5m) (a consortium led by international NGOs involved in humanitarian assistance with 20 members in Zimbabwe), who run ‘replica’ programmes paid for by international aid donors. The model will pay-out when an estimated 3.3 million people are affected (the attachment point), with a response cost of USD 154m, and so is designed for a major national disaster not for regular responses to food insecurity.
However, the ARC approach, which is an African Union initiative, with separate development and commercial arms, has had a chequered history. For example, in 2016 the insurance failed to pay out in Malawi during what was clearly on the ground a disastrous drought. In the end an ex gratia payment was made, but the approach was seriously critiqued – a damning Action Aid report argued that this was the wrong model for improving resilience. Despite its many promoters in the aid agencies, the ARC lost credibility and there was a period when it looked like it would collapse with insufficient country subscribers and too little funding to reinsure. A preliminary evaluation by OPM suggest some major flaws, particularly in the way the underlying ‘AfricaRiskView‘ model was constructed.
Since then, there have been multiple efforts at improving the system and customising the underlying model. Zimbabwe has bought into a very different operation, with the model being calibrated with local information through a committee, led by the Ministry of Finance, with many experts from the ministry of agriculture (Agritex) as well as aid agencies and NGOs. Although the details of the model are secret – they are proprietary information of the commercial arm of the ARC, and the basis on which it presumably gears a profit from its assessment of risks – there is clearly more local participation and transparency than before. Those who have premiums invested can monitor the progress through the season, assessing if a pay-out is likely.
The implementing agencies, whether government, the WFP or the START network, must come up with a contingency plan for how they will deliver assistance in case a pay-out is made. The current policy aims to reach up to 800,000 people in drought-prone districts if a full pay-out occurs. This is supposed to mean that things can happen quickly and before the worst impacts of a disaster strike.
Contingency plans currently involve the usual array of targeted interventions, with all the problems that these entail, but the principle of early action and rapid response is definitely a good one; although such plans need to be held in place and updated in all the years in between pay-outs (the ARC deal for Zimbabwe expects, but doesn’t guarantee, a one in four year pay-out; the assumed ‘return period’).
How will it work it practice? This is the first year with the latest round of insurance, so the simple answer is we don’t know. Past experience is limited in Zimbabwe, as there has only been one payment from a previous round in mid 2020 of USD 1.4 million to the government and around USD 300k to WFP (as the replica premium holder) following the poor rains of 2019-20.
As one senior official from a humanitarian agency observed, such pay-outs are all well and good, but they are a drop in the ocean. When hundreds of millions of dollars are required across over 8 million food insecure people and maybe 1.8 million in dire need, then the premiums required to cover this would be enormous, and way beyond any agency or donor, let alone the government (even though of course the ‘food insecurity’ figures may be dubious). For example, under the new policy WFP’s maximum pay-out would only be USD 6.5 million; hence people referred to the approach as a ’boutique experiment’, with varying doses of scepticism.
Other commentators we talked to were intrigued by the system, desperate to find a new way of doing things given the failures of the standard approaches but wondered whether a profit motivated company could really deliver funding for humanitarian assistance through global reinsurance markets. Should insurance brokers be making money out of disasters? For the humanitarians this was a difficult one, even if it worked (which was not sure).
There were many other debates about the practicalities and questions were still raised about the underlying model. For example, it focuses on maize as the indicator crop, but what about in areas where sorghum or other crop mixes are important? What about livestock? It was unclear if the model differentiated between different soil types – the impact of rainfall deficits is massively different between sandy and heavy soils, for example. The idea of a ‘sovereign’ drought given the variability of patterns across the country seemed odd to others. How can a single-cut off for the whole of such a diverse country be decided? And how can a single point be defined, when drought always evolves through the season?
As the next seasons unfold (the premium has been paid over several years), we will see how it pans out. A key issue for any index type insurance (where predictive risk indicators are used and actual damage is not assessed post hoc) is the question of what is called ‘basis risk’: the difference between the predictions and the actual outcomes.
If this is large (as was the case in Malawi in 2016), then the insurance pay-outs are not geared to need, and the product becomes ineffective. Those who pay the premiums or expect pay-outs rightly object and in an increasing number of cases, ex gratia pay-outs are made, effectively acknowledging the failure of the calculative risk model.
Even though the ARC model has been improved, there are plenty of reasons to expect significant basis risk, given what we know about drought in Zimbabwe (see previous blogs here and here).
From risk to uncertainty: basic challenges to insurance as a solution to disasters
There are however some more fundamental issues with insurance as a response to uncertain disasters. The challenges are more than fixing the technical parameters and reducing basis risk and improving contingency planning and implementation.
Any insurance assumes you can calculate the probabilities of an event happening (or an index being triggered). This is the basis of setting premiums where the insurance company bets on incomes versus pay-outs based on estimates of the likelihood of trigger events/disasters happening. In other words, insurance is premised on assumptions about ‘risk’ – where you have a good idea of the probability of what is going to happen (a calculable, predictive approach) – but not ‘uncertainty’ – where you don’t know the likelihood of outcomes. Under conditions of uncertainty, insurance will not work.
By assuming risk, insurance creates particular political dynamics centred on strategies of management and control. As a result, most contemporary forms of insurance are based on a modernising, market-based vision of predictive control, generating forms of ‘governmentality’ over insurance ‘subjects’. In the case of the ARC’s sovereign insurance this is scaled across a whole nation, is exerted through an insurance company with links to the global insurance markets and is executed on the ground by the state, UN agencies or international NGO networks.
This is very different to the drought responses among farmers, discussed in previous blogs in this series, with radically different politics. As discussed in the previous two blogs, it’s uncertainty not risk that dominates the experience of drought in Zimbabwe, as elsewhere. This becomes more the case as climate change affects patterns of rainfall. Taking uncertainty seriously suggests a very different approach to disaster response.
An approach centred on uncertainty, rather than risk, would have to adopt the approaches used by farmers in the face of variable patterns of rainfall and uncertain drought conditions, described in the previous blogs. This would mean flexibility, adaptability, contingency planning and performative responses to unfolding circumstances (a theme picked up in the next blog).
This is less neat than the technocratic, market solution of insurance, and for large and cumbersome agencies may be impossible. But, in the end, an uncertainty-centred approach may be more effective and more attuned to the real world of uncertainty that characterises drought and disaster, as we argued in a recent paper in relation to humanitarian and social assistance more broadly.
Time will tell how the ARC sovereign insurance model fares in Zimbabwe. Maybe it can adapt and become the flexible, early response system taking account of uncertainties that is so needed. Watch this space for updates in the coming years.