Asymmetric Information Snyder and Nicholson, Copyright ©2008 by Thomson South-Western. All rights reserved. Asymmetric Information • Transactions can involve a considerable amount of uncertainty – can lead to inefficiency when one side has better information • The side with better information is said to have private information or asymmetric information The Value of Contracts • Contractual provisions can be added in order to circumvent some of the inefficiencies associated with asymmetric information – rarely do they eliminate them Principal-Agent Model • The party who proposes the contract is called the principal • The party who decides whether or not to accept the contract and then performs under the terms of the contract is the agent – typically the party with asymmetric information Leading Models • Two models of asymmetric information – the agent’s actions affect the principal, but the principal does not observe the actions directly • called a hidden-action model or a moral hazard model – the agent has private information before signing the contract (his type) • called a hidden-type model or an adverse selection model First, Second, and Third Best • In a full-information environment, the principal could propose a contract that maximizes joint surplus – could capture all of the surplus for himself, leaving the agent just enough to make him indifferent between agreeing to the contract or not • This is called a first-best contract First, Second, and Third Best • The contract that maximizes the principal’s surplus subject to the constraint that he is less well informed than the agent is called a second-best contract • Adding further constraints leads to the third best, fourth best, etc. Hidden Actions • The principal would like the agent to take an action that maximizes their joint surplus • But, the agent’s actions may be unobservable to the principal – the agent will prefer to shirk • Contracts can mitigate shirking by tying compensation to observable outcomes Hidden Actions • Often, the principal is more concerned with outcomes than actions anyway – may as well condition the contract on outcomes Hidden Actions • The problem is that the outcome may depend in part on random factors outside of the agent’s control – tying the agent’s compensation to outcomes exposes the agent to risk – if the agent is risk averse, he may require the payment of a risk premium before he will accept the contract Owner-Manager Relationship • Suppose a firm has one representative owner and one manager – the owner offers a contract to the manager – the manager decides whether to accept the contract and what action e 0 to take • an increase in e increases the firm’s gross profit but is personally costly to the manager Owner-Manager Relationship • The firm’s gross profit is g = e + – where represents demand, cost, and other economic factors outside of the agent’s control • assume ~ (0,2) – c(e) is the manager’s personal disutility from effort • assume c’(e) > 0 and c’’(e) < 0 Owner-Manager Relationship • If s is the manager’s salary, the firm’s net profit is n = n – s • The risk-neutral owner wishes to maximize the expected value of profit E(n) = E(e + – s) = e – E(s) Owner-Manager Relationship • We will assume the manager is risk averse with a constant risk aversion parameter of A > 0 • The manager’s expected utility will be E u E s A 2 Var s c e First-Best • With full information, it is relatively easy to design an optimal salary contract – the owner can pay the manager a salary if he exerts a first-best level of effort and nothing otherwise – for the manager to accept the contract E(u) = s* - c(e*) 0 First-Best • The owner will pay the lowest salary possible [s* = c(e*)] • The owner’s net profit will be E(n) = e* - E(s*) = e* - c(e*) – at the optimum, the marginal cost of effort equals the marginal benefit Second Best • If the owner cannot observe effort, the contract cannot be conditioned on e – the owner may still induce effort if some of the manager’s salary depends on gross profit – suppose the owner offers a salary such as s(g) = a + bg – a is the fixed salary and b is the power of the incentive scheme Second Best • This relationship can be viewed as a three-stage game – owner sets the salary (choosing a and b) – the manager decides whether or not to accept the contract – the manager decides how much effort to put forth (conditional on accepting the contract) Second Best • Because the owner cannot observe e directly and the manager is risk-averse, the second-best effort will be less than the first-best effort – the risk premium adds to the owner’s cost of inducing effort First- versus Second-Best Effort The owner’s MC is higher in the second best, leading to lower effort by the manager MC in second best c’(e) + risk term MC in first best c’(e) 1 MB e e** e* Moral Hazard in Insurance • If a person is fully insured, he will have a reduced incentive to undertake precautions – may increase the likelihood of a loss occurring Moral Hazard in Insurance • The effect of insurance coverage on an individual’s precautions, which may change the likelihood or size of losses, is known as moral hazard Mathematical Model • Suppose a risk-averse individual faces the possibility of a loss (l) that will reduce his initial wealth (W0) – the probability of loss is – an individual can reduce this probability by spending more on preventive measures (e) Mathematical Model • An insurance company offers a contract involving a payment of x to the individual if a loss occurs – the premium is p • If the individual takes the coverage, his expected utility is E[u(W)] = (1-)u(W0-e-p) + ()u(W0-e-p-l+x) First-Best Insurance Contract • In the first-best case, the insurance company can perfectly monitor e – should set the terms to maximize its expected profit subject to the participation constraint • the expected utility with insurance must be at least as large as the utility without the insurance – will result in full insurance with x = l – the individual will choose the socially efficient level of precaution Second-Best Insurance Contract • Assume the insurance company cannot monitor e at all – an incentive compatibility constraint must be added • The second-best contract will typically not involve full insurance – exposing the individual to some risk induces him to take some precaution Hidden Types • In the hidden-type model, the individual has private information about an innate characteristic he cannot choose – the agent’s private information at the time of signing the contract puts him in a better position Hidden Types • The principal will try to extract as much surplus as possible from agents through clever contract design – include options targeted to every agent type Nonlinear Pricing • Consider a monopolist who sells to a consumer with private information about his own valuation for the good • The monopolist offers a nonlinear price schedule – menu of different-sized bundles at different prices – larger bundles sell for lower per-unit price Mathematical Model • Suppose a single consumer obtains surplus from consuming a bundle of q units for which he pays a total tariff of T u = v(q) – T – assume that v’(q) > 0 and v’’(q) < 0 – the consumer’s type is • H is the “high” type (with probability of ) • L is the “low” type (with probability of 1-) • 0 < L < H Mathematical Model • Suppose the monopolist has a constant average and marginal cost of c • The monopolist’s profit from selling q units is = T – cq First-Best Nonlinear Pricing • In the first-best case, the monopolist observes • At the optimum v’(q) = c – the marginal social benefit of increased quantity is equal to the marginal social cost First-Best Nonlinear Pricing This graph shows the consumers’ indifference curves (by type) and the firm’s isoprofit curves T U0H U0L q First-Best Nonlinear Pricing A is the first-best contract offered to the “high” type and B is the first-best offer to the “low” type T A U0H U0L B q Second-Best Nonlinear Pricing • Suppose the monopolist cannot observe – knows the distribution • Choosing A is no longer incentive compatible for the high type – the monopolist must reduce the high-type’s tariff Second-Best Nonlinear Pricing The “high” type can reach a higher indifference curve by choosing B T A U0H U2H C U0L B q To keep him from choosing B, the monopolist must reduce the “high” type’s tariff by offering a point like C Second-Best Nonlinear Pricing The monopolist can also alter the “low” type’s bundle to make it less attractive to the high type T A E C U0H U2H U0L B D q**L q**H q Monopoly Coffee Shop • The college has a single coffee shop – faces a marginal cost of 5 cents per ounce • The representative customer faces an equal probability of being one of two types – a coffee hound (H = 20) – a regular Joe (L = 15) • Assume v(q) = 2q0.5 First Best • Substituting such that marginal cost = marginal benefit, we get q = (/c)2 q*L = 9 q*H = 16 T*L = 90 T*H = 160 E() = 62.5 Incentive Compatibility when Types Are Hidden • The first-best pricing scheme is not incentive compatible if the monopolist cannot observe type – keeping the cup sizes the same, the price for the large cup would have to be reduced by 30 cents – the shop’s expected profit falls to 47.5 Second Best • The shop can do better by reducing the size of the small cup • The size that is second best would be LqL-0.5 = c + (H - L)qL-0.5 q**L = 4 T**L = 60 E() = 50 Adverse Selection in Insurance • Adverse selection is a problem facing insurers where the risky types are more likely to accept an insurance policy and are more expensive to serve – assume policy holders may be one of two types • H = high risk • L = low risk First Best • The insurer can observe the individual’s risk type • First best involves full insurance – different premiums are charged to each type to extract all surplus First Best W2 U0L certainty line U0H Without insurance each type finds himself at E B A and B represent full insurance A E W1 Second Best • If the insurer cannot observe type, firstbest contracts will not be incentive compatible – if the insurer offered A and B, the high-risk type would choose B – the insurer must change the coverage offered to low-risk individuals to make it unattractive to high-risk individuals First Best W2 U1H U0L certainty line U0H B The high-risk type is fully insured, but his premium is higher (than it would be at B) C D A E The low-risk type is only partially insured W1 Market Signaling • If the informed player moves first, he can “signal” his type to the other party – the low-risk individual would benefit from providing his type to insurers • he should be willing to pay the difference between his equilibrium and his first-best surplus to issue such a signal Market for Lemons • Sellers of used cars have more information on the condition of the car – but the act of offering the car for sale can serve as a signal of car quality • it must be below some threshold that would have induced the owner to keep it Market for Lemons • Suppose there is a continuum of qualities from low-quality lemons to high-quality gems – only the owner knows a car’s type • Because buyers cannot determine the quality, all used cars sell for the same price – function of average car quality Market for Lemons • A car’s owner will choose to keep a car that is in the upper end of the spectrum – reduces the average quality – reduces the market price – leads sellers of the high end of the remaining cars to keep their cars • reduces average quality and market price Adverse Selection • Consider a used car market. • Two types of cars; “lemons” and “peaches”. • Each lemon seller will accept $1,000; a buyer will pay at most $1,200. • Each peach seller will accept $2,000; a buyer will pay at most $2,400. Adverse Selection • If every buyer can tell a peach from a lemon, then lemons sell for between $1,000 and $1,200, and peaches sell for between $2,000 and $2,400. • Gains-to-trade are generated when buyers are well informed. Adverse Selection • Suppose no buyer can tell a peach from a lemon before buying. • What is the most a buyer will pay for any car? Adverse Selection • Let q be the fraction of peaches. • 1 - q is the fraction of lemons. • Expected value to a buyer of any car is at most EV $1200(1 q) $2400q. Adverse Selection • Suppose EV > $2000. • Every seller can negotiate a price between $2000 and $EV (no matter if the car is a lemon or a peach). • All sellers gain from being in the market. Adverse Selection • Suppose EV < $2000. • A peach seller cannot negotiate a price above $2000 and will exit the market. • So all buyers know that remaining sellers own lemons only. • Buyers will pay at most $1200 and only lemons are sold. Adverse Selection • Hence “too many” lemons “crowd out” the peaches from the market. • Gains-to-trade are reduced since no peaches are traded. • The presence of the lemons inflicts an external cost on buyers and peach owners. Adverse Selection • How many lemons can be in the market without crowding out the peaches? • Buyers will pay $2000 for a car only if EV $1200(1 q ) $2400q $2000 Adverse Selection • How many lemons can be in the market without crowding out the peaches? • Buyers will pay $2000 for a car only if EV $1200(1 q ) $2400q $2000 2 q . 3 • So if over one-third of all cars are lemons, then only lemons are traded. Adverse Selection • A market equilibrium in which both types of cars are traded and cannot be distinguished by the buyers is a pooling equilibrium. • A market equilibrium in which only one of the two types of cars is traded, or both are traded but can be distinguished by the buyers, is a separating equilibrium. Adverse Selection • What if there is more than two types of cars? • Suppose that car quality is Uniformly distributed between $1000 and $2000 any car that a seller values at $x is valued by a buyer at $(x+300). • Which cars will be traded? Adverse Selection The expected value of any car to a buyer is $1500 + $300 = $1800. 1000 1500 Seller values 2000 So sellers who value their cars at more than $1800 exit the market. Adverse Selection The distribution of values of cars remaining on offer 1000 1800 Seller values Adverse Selection The expected value of any remaining car to a buyer is $1400 + $300 = $1700. 1000 1400 1800 Seller values So now sellers who value their cars between $1700 and $1800 exit the market. Adverse Selection • Where does this unraveling of the market end? • Let vH be the highest seller value of any car remaining in the market. • The expected seller value of a car is 1 1 1000 v H . 2 2 Adverse Selection • So a buyer will pay at most 1 1 1000 v H 300. 2 2 • This must be the price which the seller of the highest value car remaining in the market will just accept; i.e. 1 1 1000 v H 300 v H . 2 2 Adverse Selection 1 1 1000 v H 300 v H 2 2 v H $1600. Adverse selection drives out all cars valued by sellers at more than $1600. Signaling • Adverse selection is an outcome of an informational deficiency. • What if information can be improved by high-quality sellers signaling credibly that they are high-quality? • E.g. warranties, professional credentials, references from previous clients etc. Signaling • A labor market has two types of workers; high-ability and low-ability. • A high-ability worker’s marginal product is aH. • A low-ability worker’s marginal product is aL. • aL < aH. • A fraction h of all workers are high-ability. • 1 - h is the fraction of low-ability workers. Signaling • Each worker is paid his expected marginal product. • If firms knew each worker’s type they would pay each high-ability worker wH = aH pay each low-ability worker wL = aL. Signaling • If firms cannot tell workers’ types then every worker is paid the (pooling) wage rate; i.e. the expected marginal product wP = (1 - h)aL + haH. Signaling • wP = (1 - h)aL + haH < aH, the wage rate paid when the firm knows a worker really is high-ability. • So high-ability workers have an incentive to find a credible signal. Signaling • Workers can acquire “education”. • Education costs a high-ability worker cH per unit • and costs a low-ability worker cL per unit. • cL > cH. • Suppose that education has no effect on workers’ productivities; i.e., the cost of education is a deadweight loss. Signaling • High-ability workers will acquire eH education units if (i) wH - wL = aH - aL > cHeH, and (ii) wH - wL = aH - aL < cLeH. • (i) says acquiring eH units of education benefits high-ability workers. • (ii) says acquiring eH education units hurts low-ability workers. Signaling aH aL cHeH and together require aH aL cLeH a H aL a H aL eH . cL cH Acquiring such an education level credibly signals high-ability, allowing high-ability workers to separate themselves from low-ability workers. Signaling • Q: Given that high-ability workers acquire eH units of education, how much education should low-ability workers acquire? • A: Zero. Low-ability workers will be paid wL = aL so long as they do not have eH units of education and they are still worse off if they do. Signaling • Signaling can improve information in the market. • But, total output did not change and education was costly so signaling worsened the market’s efficiency. • So improved information need not improve gains-to-trade. Auctions • A seller can often do better if several buyers compete against each other – high-value consumers are pushed to bid high • Different formats may lead to different outcomes – sellers should think carefully about how to design the auction First-Price Sealed Auction Bid • All bidders simultaneously submit secret bids • The auctioneer unseals the bids and awards the object to the highest bidder • The highest bidder pays his own bid First-Price Sealed Auction Bid • In equilibrium, it is a weakly dominated strategy to submit a bid b greater than or equal to the buyer’s valuation v – a strategy is weakly dominated if there is another strategy that does at least as well against all rivals’ strategies and strictly better against at least one First-Price Sealed Auction Bid • A buyer receives no surplus if he bids b=v no matter what his rivals bid – by bidding b < v, there is a chance for some positive surplus • Since players likely avoid weakly dominated strategies, we can expect bids to be lower then buyers’ valuations Second-Price Sealed Auction Bid • The highest bidder pays the next highest bid rather than his own • All bidding strategies are weakly dominated by the strategy of bidding exactly one’s valuation – second-price auctions induce bidders to reveal their valuations Second-Price Sealed Auction Bid • The reason that bidding one’s valuation is weakly dominant is that the winner’s bid does not affect the amount he has to pay – that depends on someone else’s bid Common Values Auctions • In complicated economic environments, different auction formats do not necessarily yield the same revenue • Suppose the good has the same value to all bidders, but they do ot know exactly what that value is – common values auction Common Values Auctions • The winning bidder realizes that every other bidder probably though the object was worth less – means that he probably overestimated the value when bidding • This is often referred to as the winner’s curse Important Points to Note: • Asymmetric information is often studied using a principal-agent model in which a principal offers a contract to an agent who has private information – the two main variants of the model are the models of hidden actions and hidden types Important Points to Note: • In a hidden-action model (called a moral hazard model), the principal tries to induce the agent to take appropriate actions by tying the agent’s payments to observable outcomes – doing so exposes the agent to random fluctuations, which is costly for a riskaverse agent Important Points to Note: • In a hidden-type model (called an adverse selection model), the principal cannot extract all of the surplus from high types because they can always gain positive surplus by pretending to be a low type – the principal will offer a menu of contracts from which different types of agents can select Important Points to Note: • In a hidden-type model, the principal will offer a menu of contracts from which different types of agents can select – the principal distorts the quantity offered to low types in order to make the contract less attractive to high types Important Points to Note: • Most of the insights gained from the basic form of a principal-agent model, in which the principal is a monopolist, carry over to the case of competing principals – the main change is that agents obtain more surplus Important Points to Note: • The lemons problem arises when sellers have private information about the quality of their goods – sellers whose goods are higher than average quality may refrain from selling – the market may collapse, with goods of only the lowest quality being offered for sale Important Points to Note: • The principal can extract more surplus from agents if several of them are pitted against one another in an auction setting – in a simple economic environment, a variety of common auction formats generate the same revenue – differences in auction format may generate different levels of revenue in more complicated settings