Notes on Information Asymmetry, Adverse Selection, and Moral Hazard in Markets

Used Cars Market: adverse selection and unraveling

• Setup

  • Buyers value cars higher than sellers by a factor of 1.5: buyers should end up with 1.5x the number of cars as sellers in a fair trade.
  • All players have money and all goods are valued similarly across other goods.
  • Each car has a quality q in [0, 100], uniformly distributed (implicitly). Sellers derive utility from keeping a car (value q in terms of other goods) plus money from selling, while buyers derive utility from acquiring cars.
  • Price p is the observed transaction price.

• Seller decision at price p

  • A seller will sell if the utility from selling (price p) exceeds the utility from keeping the car (quality q) plus other goods.
  • Given the assumed utility form (car quality + other goods, with price of other goods fixed at 1), sellers sell if the car’s quality q is less than p: q < p. Hence, at any price p, only cars with quality below p are on the market.

• Buyer decision at price p

  • A buyer does not know the quality q, only the price p. They form a conditional expectation of car quality given the price: ext{E}[q \,|\, p].
  • The buyer’s utility from buying a car (ignoring other costs for the moment) is proportional to quality, with a factor of 3/2 per unit of quality. If the price of the good is 1 per unit of other goods, the buyer’s utility from the car is rac{3}{2} q. Therefore, conditional expected utility from buying at price p is:

  {
  m EU}(p) = rac{3}{2} \mathbb{E}[q \mid p] - p.

• Distributional reasoning (the 100 cap example)

  • If p < 100, only cars with q < p are on the lot, so the conditional distribution of q is Uniform(0, p). Hence
    \mathbb{E}[q \mid p] = \frac{p}{2}, \quad p < 100.
  • Then
    {
    m EU}(p) = \frac{3}{2} \cdot \frac{p}{2} - p = \frac{3p}{4} - p = -\frac{p}{4} < 0.
  • If p = 100, all cars are on the lot (since q < p holds for all q up to 100). The conditional mean of q is then \mathbb{E}[q \mid p=100] = 50.
  • Then
    {
    m EU}(100) = \frac{3}{2} \cdot 50 - 100 = 75 - 100 = -25 < 0.
  • If p > 100, the same logic as p=100 applies (all cars are on the lot, and the expected quality is 50). The EU is even more negative: {
    m EU}(p) = 75 - p < 0.

• Conclusion for the used-car market

  • There is no price p that yields positive expected utility for buyers; hence no trade occurs in equilibrium.
  • This is called complete unraveling: the market fails to reach any trade, driven by asymmetric information: sellers know more about car quality than buyers do.
  • If both sides had symmetric information (both know the same car quality distribution), trade could be efficient; with asymmetry, the best possible outcome (selling all cars to buyers) isn’t reached.

• Health insurance analogy: adverse selection vs moral hazard

  • Flip the story: health insurers are the buyers, individuals are the sellers. Health costs are uniformly distributed; individuals know their own expected health costs, insurers do not.
  • In a simple, risk-neutral model with uniform health costs, the actuarial cost (expected payout) p equals the average health cost across the population.
  • If p is the actuarially fair premium, roughly half of individuals will not buy insurance because their own costs are below the average; the sick half buys (uniform distribution argument). This creates adverse selection: higher-risk individuals are overrepresented in the insured pool, raising the premium and driving more healthy people out, potentially leading to a death spiral.
  • This can be mitigated in some contexts by underwriting (gathering information) or government interventions; but underwriting is costly and often restricted by law.

• Adverse selection vs advantageous selection

  • Adverse selection: higher-risk individuals are more likely to purchase insurance, driving up premiums and potentially collapsing the market.
  • Advantageous selection: sometimes, healthier individuals are more likely to buy insurance, or higher-risk individuals with particular traits (risk aversion, cognitive function, etc.) are more likely to purchase certain plans, which can offset or invert the naive adverse-selection story in some contexts.
  • Example discussed: among Medicare plan enrollees, those who buy Medigap plans tend to have better cognitive function, illustrating advantageous selection.

• Role of information structure in market outcomes

  • When asymmetry exists, the market can fail to reach efficient trade even without monopolies or government price controls.
  • If everyone knew nothing (no information), the market could in principle reach a Pareto-optimal allocation, but with asymmetric information (sellers know more than buyers), trade can unravel.

• Moral hazard vs adverse selection: definitions and implications

  • Moral hazard: once insured, individuals may take more risks or consume more health care than they would otherwise because the insurer bears some of the cost.
  • Adverse selection: individuals with higher expected costs are more likely to purchase insurance, shifting the risk pool toward the high-cost segment.
  • Distinguishing them is important because the policy responses differ:
  • Moral hazard requires containment of overuse or selective coverage;
  • Adverse selection requires expanding risk pools (e.g., universal insurance) to prevent spirals.

• Identifying asymmetric information in practice

  • A simple positive-correlation test: if insured status correlates with higher costs or utilization, this suggests asymmetric information. However, such correlation can arise from either moral hazard or adverse selection.
  • To distinguish, one would ideally use an experiment that randomizes coverage (random assignment to insurance) to isolate moral hazard (causal effect of coverage on cost).
  • In practice, exact disentangling is hard; policy implications differ depending on which mechanism dominates.

• Policy responses to adverse selection and moral hazard

  • Adverse selection remedies:
  • Mandate universal insurance (everyone must participate) to pool risks and remove selection effects.
  • Subsidize insurance to lower prices and attract healthy participants, potentially creating a multiplier effect as more healthy people join.
  • Use penalties or taxes on not buying insurance to mimic a mandate.
  • Moral hazard remedies:
  • Control overutilization through tighter authorization, cost-sharing, and managed care strategies.
  • Government trade-offs
  • A mandate can reduce adverse selection but may be criticized if it imposes costs on willing non-buyers; subsidies can help, but require public funding.
  • Legal and ethical considerations
  • Underwriting can be illegal or restricted (e.g., ACA prohibits denying coverage based on health status).
  • Some forms of discrimination (by gender, race) are illegal, though health cost correlations with race or gender may exist, and policy debates revolve around these trade-offs.

• Illustrative real-world notes on regulation and underwriting

  • Insurance underwriting: collecting health histories, family history, lifestyle, etc., to assess risk and set prices; costly and sometimes illegal (e.g., cannot deny coverage for health status under ACA).
  • Disability vs health insurance premium differences: historically, pricing for disability can differ by gender due to long-term care needs, maternity-related costs, etc.; in many contexts, health insurance pricing is restricted to avoid discrimination.
  • The Affordable Care Act (ACA): imposes protections against health-status discrimination and introduces mandates/subsidies to address adverse selection; the political economy involves balancing efficiency with equity and cost containment.

• Conceptual recap: why these issues matter

  • Markets with asymmetric information can deviate from the competitive equilibrium in fundamental ways, sometimes leading to no-trade or suboptimal trade.
  • The same logic applies across domains: used cars, health insurance, and other markets where individuals know more about their own risk than buyers or sellers do.
  • Understanding whether a market failure is driven by adverse selection or moral hazard informs the policy instruments most likely to improve welfare.

• Quick problem-set pointers discussed in class

  • Expected utility with lotteries: for outcomes li with probabilities pi, the expected utility is EU = \sumi pi \cdot u(l_i).
  • Certainty equivalent (CE): the amount c such that u(c) = EU. For CRRA or quadratic utility, compute c accordingly.
  • Actuarially fair premium: p_{act} = \mathbb{E}[L], where L is the (random) payout; in a uniform distribution of costs, this is the mean cost.
  • Premium range intuition: the actuarially fair premium gives the lowest price that still clears the market under perfect competition; the upper bound is tied to the certainty equivalent or the maximal willingness-to-pay if you want full insurance; the true market premium lies somewhere in this interval depending on market structure and competition.

• Final reflections

  • The transcript emphasizes that information asymmetry can cause complete market unraveling in simple models, but real-world policies (mandates, subsidies, underwriting, legal protections) attempt to restore efficiency while balancing equity and political constraints.
  • The discussion connects to Arrow’s concerns about market failures in health insurance and foreshadows policy debates about health reform, moral hazard, and adverse selection in the healthcare system.