Asymmetric information

Most market failures discussed so far assume buyers and sellers have the same information. They rarely do. The lender knows less about the borrower than the borrower; the insurer knows less about the insured's health than the insured; the buyer of a used car knows less about its condition than the seller. These asymmetric-information markets behave differently from the textbook competitive market — sometimes catastrophically.

The Akerlof lemons model

George Akerlof (1970): suppose used cars come in two types — good ones (worth $5,000 to buyers) and bad ones (worth $2,000). Sellers know the type; buyers don't. Without information, buyers offer the average price ($3,500 if 50/50 mix). At $3,500, good-car sellers refuse to sell (they value the car at more than $3,500); only bad-car sellers transact. The market collapses to a lemons-only equilibrium where everyone knows only bad cars are being sold.

Insurance: adverse selection

Imagine an insurance company offering health insurance at premium $1,000/year. Healthy people don't bother to insure (they don't expect to use it). Sick people do (they know they'll claim). The insurer faces only sick customers; expected claims exceed premiums; the insurer loses money; either premiums rise (driving away more healthy customers) or the insurer exits. The market 'unravels'.

This is adverse selection. The customers who select INTO the insurance pool are the ones MOST likely to claim — the ones the insurer LEAST wants. The asymmetric information (customer knows their health better than insurer) creates the perverse selection.

Responses to adverse selection in insurance

• Risk-based pricing — charge sick customers more. Requires the insurer to observe health (medical underwriting). Effective but raises equity and access concerns
• Mandatory insurance — eliminate the selection problem by requiring everyone to insure. Healthy people pool with sick; the average premium is sustainable. The principle behind universal-coverage schemes including Kenya's new SHIF
• Pre-existing condition exclusions — refuse to cover conditions the customer had before buying insurance. Crude but effective in limiting adverse selection
• Group insurance — sell to groups (employers, professional associations) rather than individuals. The group includes healthy and sick mixed roughly proportional to the population
• Long waiting periods — discourage casual sign-up by people who 'know' they'll claim soon

Credit: adverse selection in lending

When banks charge high interest rates, the borrowers who self-select into the high-interest loans are disproportionately RISKY (they have low alternative options and high expected default). At a sufficiently high interest rate, the loan portfolio's expected default exceeds the interest premium, and banks lose money even on a portfolio with the high rates. The market 'unravels' similar to lemons.

Stiglitz and Weiss (1981) formalised this. The implication: banks may NOT raise interest rates to clear demand. Instead, they ration credit at sub-equilibrium rates — turning down some borrowers entirely rather than charging them risk-appropriate rates. Credit rationing is a rational response to adverse selection.

Moral hazard

Distinct from adverse selection. Moral hazard arises AFTER the contract is signed: the insured / borrower / employee changes behaviour in ways that the contracting party can't observe (or can only observe imperfectly). Examples:

• Insured drivers drive more carelessly because the insurer covers the loss
• Insured patients seek more healthcare than they would if uninsured (the 'flat-of-the-curve' problem)
• Borrowers take on riskier projects when their downside is limited (limited liability + corporate veil)
• Employees work less hard when output isn't directly observable

Responses to moral hazard

• Co-insurance and deductibles — customer pays a portion of any claim. Aligns customer incentives with insurer incentives
• Monitoring — observe behaviour where possible. Telematics for insurance, regular performance reviews for employees
• Performance pay — link compensation to observable output. Bonus structures, sales commissions
• Collateral — borrower posts assets; bank seizes them on default. Aligns borrower's incentive to repay

Signalling and screening

Two mechanisms for getting around information asymmetry:

• Signalling — the INFORMED party takes a costly action that demonstrates their type. Education is the canonical example (Spence 1973). A worker takes more education even though education doesn't add to productivity — because employers can observe it and use it as a signal of productivity. The cost of education is high for less-productive workers (the signal is costly to fake)
• Screening — the UNINFORMED party offers a menu of contracts; the informed party self-selects (Rothschild-Stiglitz 1976). Insurance companies offer high-premium-low-deductible vs low-premium-high-deductible plans; risk-averse customers (more likely to claim) choose the first, low-risk customers the second. The menu separates types via self-selection

Microfinance — group lending as a response to information asymmetry

The classic Grameen Bank model (Yunus 1976+) addressed credit market asymmetric information through:

• Group lending — borrowers form a group; the group is collectively responsible for repayment. Borrowers know each other better than the bank does; they self-select for creditworthiness; they monitor each other; they pressure non-payers. Effectively, the group internalises the information asymmetry
• Step-by-step lending — start with small loans; successful repayment unlocks larger loans. The borrower's revealed behaviour (repayment) reduces information asymmetry over time
• Regular meetings — weekly or biweekly group meetings build social capital and reduce default through peer pressure
• Female focus — Grameen's emphasis on women borrowers reflected empirical observation that women had better repayment than men in the target population. Whether this was because women are more responsible or because women had more limited outside options is debated

The group-lending model achieved repayment rates of 95%+ in many programmes — vastly better than conventional credit-market performance with similar borrowers. The mechanism is the joint-liability resolution of the information asymmetry.

Used-goods markets in Africa

Used-clothing (mitumba), used vehicles, second-hand electronics — all asymmetric-information markets. Akerlof's lemons concerns apply. Adaptations:

• Used cars — extensive verification rituals (mechanic inspections, history-check services, third-party warranties). The standard Kenyan used-car purchase involves a pre-purchase mechanic inspection — a private response to information asymmetry
• Mobile phones — local repair markets exist; sellers offer test periods; warranty periods provide some signal of seller confidence
• Real estate — title searches, surveyor reports, property history checks. Title fraud is a known issue and constrains market efficiency