Risk and insurance

Risk is everywhere in African household life — health shocks, drought, livestock loss, business failure, theft, family deaths. Formal insurance penetration is low (Africa-wide insurance density is about 0.5% of GDP vs 7-10% in OECD). The puzzle: why isn't insurance more widely demanded? The answer combines standard risk-aversion with behavioural and institutional features that the textbook insurance model misses.

Expected-utility risk aversion (standard)

The textbook treatment: agents have a concave utility function over wealth. A concave function means equal-sized losses hurt more than equal-sized gains help. The certainty equivalent of a gamble is less than its expected value. The agent will pay a premium (the risk premium) to avoid the gamble. Insurance is the market mechanism that supplies this premium.

The behavioural twist

Prospect theory's loss aversion is empirically much stronger than expected-utility-theory risk aversion alone. The kink at the reference point creates strong aversion to losses specifically — not just to variance per se.

• Pure risk aversion would imply willingness to buy insurance against all variance — both upside and downside. People don't insure against upside (lottery wins) and don't even want to. Risk-aversion alone is wrong
• Loss aversion implies strong willingness to avoid downside losses, but with the asymmetry creating different behaviour from pure risk-aversion. Empirical insurance demand patterns match prospect theory better than expected-utility theory

Why insurance under-purchasing in African contexts?

Despite high underlying risk and strong loss-aversion, formal insurance is under-purchased. Multiple mechanisms:

• Probability weighting (prospect theory) — small probabilities are over-weighted, but VERY small probabilities are weighted at zero. Household risk of cyclone or earthquake in some regions is weighted at zero in mental computation. Insurance against rare events feels like wasted money
• Reference-point shifting — if the reference point is 'today's normal,' a loss is salient. If the reference point gradually shifts (climate adaptation), the loss feels less acute and insurance demand falls. In drought-prone regions, loss frequency is high but the reference-point also adapts
• Liquidity constraints — the premium has to be paid up front in cash. Households short on cash can't afford the premium even when the protection would be valuable
• Limited trust — formal insurance requires trust that the insurer will pay claims. African insurance markets have history of denied claims, slow payouts, regulatory inconsistency. Trust is justifiably low
• Subsistence economics — when current consumption is near subsistence, paying an insurance premium reduces today's consumption to below subsistence in exchange for protection against future tail risk. The mathematics says don't insure
• Informal alternatives — extended family, religious community, chama, neighbours all provide insurance-like risk-pooling. Formal insurance competes with these informal alternatives
• Mental accounting — insurance premium is mentally categorised as 'money out the door' rather than 'reduced expected loss'. The accounting penalises insurance more than is warranted

Informal insurance mechanisms

Where formal insurance is under-supplied, informal mechanisms take over:

• Extended family support — sending remittances when a family member needs help; receiving them when you need help. Reciprocal arrangement that effectively pools risk across the family network
• Chama and rotating savings groups — many chamas include emergency-loan provisions; members can access funds when shock hits
• Religious community — church congregations, mosque communities, ethnic associations all provide help during family emergencies (funeral costs, medical bills, school fees)
• Hyper-local credit — local shopkeepers, market vendors give credit during hardship, expecting repayment when income flows
• Asset diversification as insurance — owning livestock, holding land, having multiple income sources spreads risk across asset classes

Microinsurance product design

Microinsurance is the small-scale insurance designed for low-income markets — low premiums, small coverage, simple products. Examples in Kenya:

• M-TIBA / Kenya Health Insurance Fund — health-coverage products distributed through M-PESA, premiums starting at KES 200/month
• Index-based weather insurance — Kilimo Salama (now Acre Africa) — agricultural insurance tied to rainfall data rather than individual crop losses. Premium triggered automatically by rainfall index dropping below threshold
• Funeral/last-expense insurance — Britam, Old Mutual products covering burial and last-expense costs. Premiums modest; payout structured for specific cultural needs
• Mobile-money-linked insurance — Airtel Insurance, Safaricom Insurance — auto-renewing low-cost protection bundled with mobile-money use

What works in microinsurance design

• Low and predictable premiums — KES 100-500/month range. Premiums above this rate hit liquidity constraints
• Index-based triggers — for weather, the payout is automatic when rainfall index drops. Reduces verification cost and dispute risk. Customer trust higher
• Distribution through mobile money — leverages existing trust in M-PESA, reduces transaction costs
• Bundling — package insurance with another product (loan, current account) so it's not a separate decision. Customer enrols by default
• Simple language — avoid actuarial jargon. Explain in 1-2 sentences what's covered, what's not, what triggers payout
• Visible payouts — claim payouts in cash via M-PESA, with confirmation. Trust requires public, fast payout

What doesn't work

• Complex products with long waiting periods — customers don't trust them
• Expensive individual underwriting — costs too much relative to small premiums
• Slow claims processing — kills trust
• Premium increases mid-contract — feels like betrayal
• Exclusions that activate at claim time — customers don't know about them when they buy, and find out when they need to claim. The single largest cause of insurance-trust failure

Behavioural-informed insurance policy

Lessons from the empirical record:

• Public-sector role — for high-loss-aversion risk (health, agricultural climate), the welfare gain from coverage exceeds the willingness-to-pay because of behavioural under-demand. Public partial-subsidy is economically justified to bridge this gap. Kenya's SHIF takes this approach for health
• Automatic enrolment — for risks that affect all households similarly (health, basic life), default enrolment with opt-out captures the loss-aversion lock-in. SHIF default enrolment is the operationalisation
• Mobile-money distribution — the distribution channel matters more than the actuarial design at the low-income margin. Trust in the channel substitutes for trust in the insurer
• Subsidies for the very poor — for households at or below subsistence, subsidised premiums are essentially the only way they can access insurance. Should be public-subsidy financed