markets with asymmetric information

29
Markets with Asymmetric Information Pasquale Schiraldi LSE February 28, 2009 PS (Institute) Markets with Asymmetric Information February 28, 2009 1 / 28

Upload: fordlovers

Post on 20-Jan-2015

583 views

Category:

Documents


4 download

DESCRIPTION

 

TRANSCRIPT

Page 1: Markets with Asymmetric Information

Markets with Asymmetric Information

Pasquale Schiraldi

LSE

February 28, 2009

PS (Institute) Markets with Asymmetric Information February 28, 2009 1 / 28

Page 2: Markets with Asymmetric Information

Markets with Asymmetric Information

Agents in a market have private information regarding their own

characteristics: �Hidden Information�

willingness to paycostquality of product

behavior: �Hidden Action�

secret price cute¤ort

PS (Institute) Markets with Asymmetric Information February 28, 2009 2 / 28

Page 3: Markets with Asymmetric Information

Markets with Asymmetric Information

General theme:

private information can lead to ine¢ ciencypossessors of private information can exploit this: �information rents�

Key areas

Contract Theory: study how contracts can/should be written toovercome problems created by asymmetric informationMechanism Design: study of how to design the rules for interaction ina market to overcome problemsMarkets for experience goods: quality of good not observable toconsumers. Consumers can learn quality by consumingAuctions: natural way to organize a market when demand is uncertain

PS (Institute) Markets with Asymmetric Information February 28, 2009 3 / 28

Page 4: Markets with Asymmetric Information

Suppose an object with value v � U [0, 1] is o¤ered by a sellerv is privately observed by the seller

Seller�s valuation:us = θsv

Buyer�s valuation:ub = θbv

whereθb > θs

=) trade is always socially optimal

PS (Institute) Markets with Asymmetric Information February 28, 2009 4 / 28

Page 5: Markets with Asymmetric Information

At price p seller would sell if

θsv � p

By selling, the seller signals that

v � pθs

(*)

The buyer buys the object if

θbE [v ] � p

Since the buyer knows (*), she can form a conditional expectation:

θbE [v jv �pθs] � p () θb

p2θs

� p

PS (Institute) Markets with Asymmetric Information February 28, 2009 5 / 28

Page 6: Markets with Asymmetric Information

Thus, for sale to occurθb � 2θs

Unless the tastes di¤er substantially, the market breaks downcompletely

Notes

market mechanism in which a lower price increases sales fails to work.Lowering the price decreases average qualitymarket may not disappear, but display a less than optimal transactionvolume.one party to transaction has private informationonly certain types of the informed party will agree: The least desirabletypes: �Adverse Selection�

Other examples:

Houses: buyer�s remorseHiringInsuranceDating

PS (Institute) Markets with Asymmetric Information February 28, 2009 6 / 28

Page 7: Markets with Asymmetric Information

Studies of the Used Car Market

Hard to measure information asymmetries

the best people have done sofar:

test for presence of information asymmetry

Empirical research focuses on used car market

important market: 11 million used vehicles sold by new car dealersalone (1994)examples are Bond(1982,1984), Lacko (1984)

PS (Institute) Markets with Asymmetric Information February 28, 2009 7 / 28

Page 8: Markets with Asymmetric Information

Suppose you observe the repair expenditures on used cars repairsi andwhether they have been traded or not.

Consider the following regression:

repairsi = α+ Xi β+ δ1fi=tradedg + εi

where Xi is a vector of vehicle characteristics.

Empirical evidence sofar inconclusive

Some empirical evidence in Genesove (JPE 1993): Paper providesevidence for adverse selection, but unable to quantify it.

Alternative explanation: �E¢ cient sorting�

PS (Institute) Markets with Asymmetric Information February 28, 2009 8 / 28

Page 9: Markets with Asymmetric Information

Hendel & Lizzeri (AER 1999)

look at dynamic model

consumers long lived, di¤er in valuation for quality

constant supply of new cars

cars last for two periods and depreciate

owner observes quality and decides whether to sell

high valuation customers buy new cars, low valuation customers buyold cars

=) new source of gains from trade: people sell used car because theyprefer new ones

used car market never shuts down: lowest valuation consumer whoowns a new car is indi¤erent between selling or keeping the car,everybod with a higher valuation sells.

PS (Institute) Markets with Asymmetric Information February 28, 2009 9 / 28

Page 10: Markets with Asymmetric Information

Continuum of in�nite-lived consumers with valuation for qualityθ 2 [0, 1], F (.)The semidurable good is �new�during the period in which it isproduced, �used�during the following period and worthlessthereafter. Durability is associated with the quality of used goods.

The quality of new units is normalized to v and the quality of usedunits is denoted by w 2 (0, 1).Stationary enviroment with NO transaction costs

PS (Institute) Markets with Asymmetric Information February 28, 2009 10 / 28

Page 11: Markets with Asymmetric Information

New Cars

Used Cars

PS (Institute) Markets with Asymmetric Information February 28, 2009 11 / 28

Page 12: Markets with Asymmetric Information

Transaction costs

By assuming that if a used car changes hands a transaction cost T isincurred, the transaction cost generates incentives to keep a used car whilethe deterioration of quality generates the gains from trade. It is easy to seethat the number of consumers who keep their used car is increasing in thesize of the transaction cost.

PS (Institute) Markets with Asymmetric Information February 28, 2009 12 / 28

Page 13: Markets with Asymmetric Information

Hendel & Lizzeri (AER 1999)

always positive volume of trade in the secodary market with adverseselection.

A type θ consumer sells used cars of quality below x(θ) and keepsused cars with quality above x(θ).

x(θ) is increasing in θ, it means that higher types on average sellhigher-quality cars and consume higher quality.

e¤ect of adverse selection on prices:

lowers used car pricese¤ect on new car prices ambiguous

marginal type buying new cars is lower (! lower price)option value of keeping the car increasessubstitution e¤ect: used car becomes worse alternativeresale value e¤ect goes down

PS (Institute) Markets with Asymmetric Information February 28, 2009 13 / 28

Page 14: Markets with Asymmetric Information

Hendel & Lizzeri (AER 1999): Implications with AdverseSelection

Consider two brands Fords and Hondas with the same quality if new.

Suppose Hondas are more reliable then Fords: example sameexpected value for the used cars, but no lower (zero) standarddeviation for Hondas.

Under adverse selection the price decline is steeper and the volume oftrade is lower for the unreliable cars. Furthermore, the set ofconsumers is segmented as follows:

PS (Institute) Markets with Asymmetric Information February 28, 2009 14 / 28

Page 15: Markets with Asymmetric Information

Hendel & Lizzeri (AER 1999): Implications with AdverseSelection

in general, the volume of trade in used Fords is less than 100 percentsince used Fords will typically be subject to adverse selection

new Ford buyers sometimes keep their used cars and new Hondabuyers never do so, Honda buyers consume a higher quality onaverage over their lifetime. Thus, new Honda buyers are higher types.

PS (Institute) Markets with Asymmetric Information February 28, 2009 15 / 28

Page 16: Markets with Asymmetric Information

Hendel & Lizzeri (AER 1999): Implications with AdverseSelection

if the rental price of Hondas is higher than that of Fords nobodywould be willing to buy Hondas.Ford buyers get the option of keeping the high realizations of qualityof the used. Honda buyers do not get this option. Thus, the rentalprice of a Honda must be smaller.It seems unintuitive that a consumer would be willing to pay a higherrental price for a less reliable car (when they have the same meanqualities).Why wouldn�t all consumers buy the more reliable car after all it givesthe same �rst-period quality at a lower cost.The intuition behind the result is that Fords appeal to a segment ofconsumers who, at equilibrium prices, are not willing to pay enough toconsume quality v every period. By buying new Fords, and keepingthem some of the time, they consume an average quality lower atlower cost.PS (Institute) Markets with Asymmetric Information February 28, 2009 16 / 28

Page 17: Markets with Asymmetric Information

Hendel & Lizzeri (AER 1999): Implications withoutAdverse Selection

The steeper price decline of a Ford relative to a Honda can also beexplained by a faster Ford quality deteriorationAssume that VF > VH> WH > WF

Since there is no asymmetric information, in the absence oftransaction costs, there would be 100- percent trade in the usedmarket.

If there are small transaction costs there will be no 100- percent tradeof Hondas. The price decline is steeper and the volume of trade islarger for Fords.PS (Institute) Markets with Asymmetric Information February 28, 2009 17 / 28

Page 18: Markets with Asymmetric Information

Hendel & Lizzeri (AER 1999)

We have shown that both models lead to the prediction of steeper pricedeclines for Fords. However, they lead to opposite predictions on thevolume of trade in the used market

PS (Institute) Markets with Asymmetric Information February 28, 2009 18 / 28

Page 19: Markets with Asymmetric Information

Lemon and Lease in the Used Business Aircraft Market(Gilligan, 2004)

Consumers with a strong preference for quality wish to frequentlyreplace durables that deteriorate rapidly

With full information, prices re�ect this rapid deterioration, andbrands with high trading volumes depreciate more quickly.

When deterioration is random and known only to sellers, onlylow-quality goods are traded at reduced prices and brands with lowtrading volumes depreciate more quickly.

PS (Institute) Markets with Asymmetric Information February 28, 2009 19 / 28

Page 20: Markets with Asymmetric Information

Hendel & Lizzeri (2002)

Leasing contracts mitigate adverse selection by increasing the supplyof used durables for sale without regard to quality

promote the positive association between trading volume anddepreciation.

PS (Institute) Markets with Asymmetric Information February 28, 2009 20 / 28

Page 21: Markets with Asymmetric Information

Gilligan, 2004

Under complete information, the quality deterioration is directlyrelated to the durable good�s price decline

when quality deterioration is small relative to transaction costs

some buyers will retain their unit rather than incur the costs associatedwith transactions of used durable goods.

when quality deterioration is large relative to transaction costs

more owners of the durable good wish to sell their units and purchase anew one:the trade volume is related to the quality depreciation

PS (Institute) Markets with Asymmetric Information February 28, 2009 21 / 28

Page 22: Markets with Asymmetric Information

Gilligan, 2004

Consider two brands of a durable good, brand X and brand Y, suchthat vx > vy > wy > wx .

Brand X deteriorates more than brand Y.

the price decline is larger and the volume of trade is greater for brandX than for brand Y.

PS (Institute) Markets with Asymmetric Information February 28, 2009 22 / 28

Page 23: Markets with Asymmetric Information

Under asymmetric information, the quality of the used durable goodis a random variable whose realization is known only to the seller andnot to potential buyers.

Consider two brands of a durable good, brand R (for reliable) andbrand U (for unreliable), such that

vr = vu , w lr = whr = E (wr ) and w

lu < E (wu) < w

hu

price decline is greater and the volume of trade is smaller for theunreliable brand.

PS (Institute) Markets with Asymmetric Information February 28, 2009 23 / 28

Page 24: Markets with Asymmetric Information

Under leasing policy

the returned units are of higher average quality than units sold in theused market under selling contractstherefore price declines are lower under leasing.

Both of these e¤ects indicate that leasing ameliorates theconsequences of adverse selection in durable goods markets; that is,leasing mitigates the reductions in both trading volume and usedprices caused by adverse selection.

PS (Institute) Markets with Asymmetric Information February 28, 2009 24 / 28

Page 25: Markets with Asymmetric Information

Gilligan, 2004

Let r be the depreciation rate, where r = f (V ,Q, L)

where V =volume of trade, Q =quality uncertainty and L =leasingfrequency

Under complete information we must have that f1 > 0 and f12 =f13 = 0.

Under asymmetric information instead f12 < 0 and f13 > 0

f12 < 0 : all else equal, aircraft brands with higher levels of qualitativeuncertainty (i.e., larger values of the cumulative number ofairworthiness directives variables) exhibit a more negative relationshipbetween trading volume and estimated depreciation

f13 > 0 : All else equal, aircraft models with higher leasing frequenciesexhibit a more positive relationship between trading volume anddepreciation rates.

PS (Institute) Markets with Asymmetric Information February 28, 2009 25 / 28

Page 26: Markets with Asymmetric Information

Gilligan, 2004

They test the above implication using U.S. data of business aircraftfrom 1980-1999.

They use the following regression:

log(P iτtP iττ

) = log(Aiτt )� Age iτt [α0 + α1V iτt + α2Q iτt + α3Liτt

+α4�V iτt

�2+ α5

�Q iτt

�2+ α6

�Liτt�2+

αα7V iτt Qiτt + α8V iτt L

iτt + α9Q iτt L

iτt + µiτt ] + εiτt

Under complete information α1 + 2α4V iτt > 0, and α7 = α8 = 0.

Asymmetric information implies α7 < 0, α8 > 0

Results are also rubost to endoneity of V ,Q and L.

PS (Institute) Markets with Asymmetric Information February 28, 2009 26 / 28

Page 27: Markets with Asymmetric Information

TABLE 3Single-Equation Estimations of the Aircraft Residual Value Function

Dependent Variable: Natural Logarithm of Aircraft Real Residual Value

Independent Variable (3.1) (3.2) (3.3) (3.4)

Constant .0392(15.26)

.0600(5.76)

.0596(5.65)

.0542(4.30)

Aircraft age �.0004(1.13)

�.0004(1.09)

�.0004(1.67)

Model fleet size �2.72#10�5

(1.35)�2.76#10�5

(1.39)�6.19#10�5

(5.33)Brand fleet size �1.67#10�4

(1.81)�1.68#10�4

(1.80)�1.53#10�4

(3.18)Aircraft type (jetp1) .0052

(.97).0053

(.97).0064

(1.14)Volume of trade .0021

(1.11).0068

(1.26)Cumulative number of airworthi-

ness directives.0002

(.30)Number of aircraft operated un-

der leasing contract1.07#10�5

(.06)Squared volume of trade �1.81#10�5

(.03)Squared cumulative airworthiness

directives�9.21#10�6

(.55)Squared number of leased aircraft 1.06#10�6

(.64)Interaction of volume of trade

and cumulative airworthinessdirectives

�.0012(2.08)

Interaction of volume of tradeand number of lease-operatedaircraft

.0006(2.26)

Interaction of cumulative airwor-thiness directives and numberof leased operated aircraft

5.78#10�6

(.59)

Effect of qualitative uncertainty onaircraft real residual valuea

�.0167(2.48)

Observations 364/5,410 364/5,410 364/5,409 364/5,406Adjusted 2R .8071 .8295 .8298 .8528Exclusion restrictions:

Cumulative airworthinessdirectives 2.29*

Number of lease-operatedaircraft 4.75**

Both cumulative directives andlease-operated aircraft 3.16**

Note.—Following eq. (1) in the text, this table reports the results of estimating the natural logarithm of real residualvalue as a function of year and brand fixed effects, the cumulative number of airworthiness directives, and the negativeof aircraft age interacted with the listed variables. Weighted (by age) and clustered (by model) standard errors areemployed. The numbers reported in parentheses below the parameter estimates are t-statistics. The F-statistics are testsof the hypotheses that qualitative uncertainty and leasing frequency are, individually and jointly, irrelevant for explainingvariation in aircraft depreciation rates.

a This row reports the parameter estimate and t-statistic for the direct effect (i.e., not interacted with aircraft age) ofthe cumulative number of airworthiness directives variable on residual value. Recall that a negative coefficient is predictedprovided that the cumulative number of airworthiness directives is a valid proxy of either qualitative uncertainty oraircraft quality.

* .p ! .05** .p ! .01

Page 28: Markets with Asymmetric Information

Estimated depreciation rate with respect to trading volume across rangesof quality uncertainty and number of lease-operated aircraft.

PS (Institute) Markets with Asymmetric Information February 28, 2009 27 / 28

Page 29: Markets with Asymmetric Information

This elasticity is the derivative of the computed depreciation rate withrespect to trading volume

Inspection of the Figure indicates that the elasticity ranges in waysconsistent with theory:

it achieves its highest value for low levels of qualitative uncertainty andhigh leasing frequencies.It takes on its lowest, indeed negative, values for high qualitativeuncertainty and low leasing frequencies.

PS (Institute) Markets with Asymmetric Information February 28, 2009 28 / 28