Introduction to Mathematical Economics
Introduction to Mathematical Economics ECON C103
Popular in Course
Popular in Economcs
This 28 page Class Notes was uploaded by Dr. Janiya Bernier on Thursday October 22, 2015. The Class Notes belongs to ECON C103 at University of California - Berkeley taught by D. Sraer in Fall. Since its upload, it has received 31 views. For similar materials see /class/226706/econ-c103-university-of-california-berkeley in Economcs at University of California - Berkeley.
Reviews for Introduction to Mathematical Economics
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 10/22/15
Econ 103C Lecture 6 An introduction to Auction Theory David Sraer Berkeley U n iversity April 21 2008 Some common auction forms a Open ascending price auction or English auction 0 Open descending price auction or Dutch auction not commonly used in practice 0 Sealed bid first price auction 0 Sealed bid second price auction a ebay 3 7 i I Aprii 2L 2008 2 30 Valuations o Auctions are used because the seller ignores the values that bidders attach to the object a Question design a mechanism to overcome this informational issue when only 1 item is sold 9 Two situations 0 The bidder knows the value of the object to himself at the time of bidding A private values EX a CD on ebay 9 Each bidder only have an imperfect value of the true common value of the object A common values EX bidding for an oil field April 21 2008 3 30 Evaluating auctions Two grounds on which to evaluate auctions 0 Revenue How much utility does a particular auction provide the seller of the good with o Efficiency does the object fall in the hand of the agent with the highest utility for it 3 s ADVll at 2008 4 30 What is an auction Two important general characteristics of auctions o Auctions are mechanisms that elicit information in the form of bids from potential buyers of a good regarding their willingness to pay and the outcome is determined solely on the basis of the received information a universality o In an auction the identity of the bidder plays no role in determining who wins the object and who pays how much a anonymity April 21 2008 5 30 The formal framework 1 0 There is a single object for sale 0 There are N potential buyersbidders 0 Each bidder i E 17 N is willing to pay X for the object at most 0 Each X corresponds to the realization of a random variable X which is iid on some interval 07w according to the increasing distribution function F continuous and with full support 0 EX lt 00 April 2L 2008 6 3o The formal framework 2 a Bidder 139 knows only the realization X of his own valuation and simply knows that others39 valuation are drawn iid from F o Bidders are risk neutral o All components of the model other than the realized values are commonly known to all bidders 0 There are no liquiditybudget constraints each bidder can pay up to hisher value X April21 2008 1 3o Analysis of second price auctions 1 0 Call 3 07w HR the equilibrium bid function of bidder i 0 An equilibrium is defined by an n tuple of bidding functions 3 73 such that for all bidder i with value X it is optimal to bid 3Xi if other bidders use the bid functions B 0 We are interested in symmetric equilibria ie in equilibria where all bidders use the same bid function BA 0 Consider a second price auction Call b the bid of player 139 His payoff is given by X 7 max by if b gt max by 7 J79 JI I 0 if b lt max by h I ll April 21 2008 s 30 Analysis of second price auctions 2 Proposition In a second price sealed bid auction it is a weakly dominant strategy to bid according to 3X X 0 Assume the other player 7i bid according to 3 Call p1 maxht bj maxhtixj the highest competing bid 0 Assume that bidder i decides to bid 2 lt X If p1 lt z or p1 gt Xi then bidding z or bidding X is equivalent However if X gt p1 gt 2 then bidding X strictly dominates bidding z 0 Assume similarly that bidder i decides to bid 2 gt X If p1 lt X or p1 gt z then bidding z or bidding X is equivalent However if z gt p1 gt X then bidding X strictly dominates bidding z Remark The previous proof does not rely on the iid assumption 39 quot 39 7W4 39ll April 21 2008 9 30 Expected payment in a second price auction 0 In a second price auction 0 A bidder i with valuation x wins the auction with probability Cx where G is the distribution function of YllNill defines as the highest value among N 71 bidders 9 Conditional on winning the auction bidder i will pay on expectation EMN UMN lt x a Note that G can be easily computed PY1quot 1lt x PX1 lt x x1 lt x FXN1 0 Overall a bidder with valuation X expected payment is given by FXN 1EY1N 1lY1N 1 lt X April 21 2008 3910 3o First Price Auctions 0 In such an auction the payoff for bidder i submitting a sealed bid b is given by X 7 b b gt Hi J79 0 if b lt maxbj J79 0 Clearly the previous equilibrium strategy is no longer optimal as it would deliver 0 profit to the bidder for sure 0 Clear tradeoff for bidder 139 increasing the bid 1 increases the probability of winning but 2 decreases the expected gain conditional on winning April 21 2008 3911 30 First Price Auctions 2 0 Assume there is a symmetric increasing and differentiable equilibrium strategy B 0 Consider bidder i with valuation X and assume the other bidders play the equilibrium strategy 3 First notice that it is never optimal for bidder i to choose a bid b gt w as reducing the bid would not affect the probability of winning the auction while increasing profit conditional on winning 0 A bidder with value 0 will bid 0 ie 0 0 0 Bidder 139 wins the auction whenever b gt maxiigl BX gt Since 3 is increasingmaX1 BX max1Xi y1 v 1 where Y1N71 is defined as before I ll April 21 2008 12 30 Firs Price Auctions 3 0 Thus bidder 1 wins the auction whenever WM lt b gt YE lt W 1b 0 His expected payoff by bidding b is therefore G lt 71 X7 b where G is again the distribution of VIM71 0 Take the FOC with respect to b g 3 71 b lt 71 gt xi b 7 G WWW 0 W W bi 0 Under a symmetric equilibrium the optimal bid b must be BMW ie 500 WW X gX X XgX o Equivalently d a CX x XgX 0 Using 30 0 we have mx foxygydy EMN IMN lt 1 I H Aprii 21 2008 13 30 First Price Auctions 4 o The equilibrium bidding strategy B can be rewritten using integration by parts G BWX WM 7 Kandy xi f3dy 0 Obviously the equilibrium bids are lower than the value X s m 1 0 As GM 7 RX the degree of shading depends on the number of competing bidders as N increases the equilibrium bid approaches X I April 21 2008 3914 30 Two exa m ples 0 Assume that the values are uniformly distributed on 01 0 If FX X then CX X V 1 and BMW X 7 f ll71 dy x a the bidder bids a constant fraction of his value a j 39 Anni 211008 3915 no Revenue comparison Proposition With i i d private values the expected seller39s revenue in a first price auction is the same as in a second price auction 0 A bidder with value X in a first price auction will win with probability Cx a Conditional on winning the auction he will pay EYfquot 1l Yf V l lt x o The expected payment by a bidder with value X is thus the same in a first price and in a second price auction 0 The ex ante payment of a particular bidder is given in both auctions by EmX fsd mXfxdx where W CXEY1N 1lY1N 1 lt X I ll April 21 2008 16 30 Revenue comparison EmX o The ex ante expected revenue generated by the seller is R N x fowy i Fygydy 0 Call YZW the second highest of N values We have 1PY2N k 1PY1N k NPY1N gt kamp Y1quot 1 lt k FkN N 1 7 Fk FkN 1 NFkN 1 7N71Fkquot I April 21 2008 11 30 Revenue comparison 0 Thus the density associated with YQN is awn NmeivMVMW4ueFM 0 But gk N 71fkFkN 2 so that QWMMMWHW 0 Thus R A f2NydyEY2Ni o The seller39s revenue is equivalent in the second and the first price auction Revenue equivalence resut I Aprii 21 2008 3918 30 Revenue comparison variance Proposition With iid private values the distribution ofequilibrium prices in a second price auction is riskier in the sense ofa mean preserving spread than the distribution of equilibrium prices in a first price auction 0 Let39s first define a mean preserving spread Let X be a random variable with distribution F 0 Let Z be a random variable such that for all potential values X of X EZlX X 0 and call H the distribution of Z a Call Y X l Z the random variable obtained from drawing from X and then drawing Z from the conditional distribution on X ie HlX X Call G the distribution of Y 0 G is a mean preserving spread of F a Y and X have the same mean but Y is more spread out than X as it isjust X l some noise 39l April 21 2008 19 30 I Revenue Comparison Variance 0 Remember that 1 in a second price auction the seller39s revenue is directly the random variable R2 YZW and 2 in a first price auction it is given by the random variable R1 BY1N o ElelR1 p EY2Nl Y1N p ElelNllYfN W p 0 But quite simply for all y m M y EiYEN WYfN lt y o So finally ER2lR1 p EY1N 1lY1N 1 lt ylt1p w 1p p 0 Thus if one call Z the random variable R2 7 R1 we have 1 R2 R1 l Z and 2 EZlR1 p 0 Thus R2 is a mean preserving spread of R1 I ll April 21 2008 20 3o Reserve price 0 O O 0 The seller often uses the possibility to set a reserve price r gt 0 ie a price that the winner of the auction will have to pay if the price set in the auction is lower than r If no bid exceeds the reserve price then the good is left unsold Consider the case of a second price auction The equilibrium strategy for a bidder is still to bid its value the reserve price excludes bidder with value lt r but the proof derived above can be applied to bidders with value 2 r The expected payment of a bidder with value equal to r is simply rGr The expected payment of a bidder with value X 2 r is 60 ffygmdy ill April 21 2008 2130 Reserve price first price auction 0 In a first price auction the reserve price still excludes bidder with value X lt r a A bidder with value r has to bid BU r 0 Now we can apply the exact same proof as before to find the following differential equation d a CX x XgX 0 So that 6X frxngMy o The expected payment of a bidder with value X 2 r is thus given by X mx7 r Gmmx ram ygwy gt The revenue equivalence still holds when there is a reserve price April 21 2008 22 3o Reserve price effect on revenue 0 We can compute the ex ante expected payment of a bidder in both the second and the first price auction as mm w ram Dandy fxdx rGr1e Fr Wendy fxdx m rGr1e Frl1 PM wow V r 0 xgx1eFxdx a Call X0 the seller39s utility if the good is left unsold The seller39s utility is thus I39lo Nmr l FrNX0 0 Utility maximizing reserve price dr N1 PM 7 rl r Cr Crfrxo I a 77 April 21 2008 23 3o Reserve price effect on revenue 2 9 Introducing the hazard rate X 1220 dl39l Tr N e 0 7x0 m 1 e Fm cm 0 In r X0 the derivative is positive a the optimal reserve price is more than the seller39s value for the good 0 Why Consider X0 O and consider setting r gt O assume only 2 bidders Tradeoff 0 Possibility that the good is left unsold Expected probability Fr2 Magnitude of loss r 9 Increase in the price paid for the good when Y2 lt r Expected probability 2Fr17 Fr Magnitude of gain r a When r is close to 0 first order gain vs second order loss gt strict gain by imposing a gt O reserve price It is optimal to exclude some bidders with value larger than X0 in order to raise the expected price exclusion principle I W ill Apl ll 21 2008 24 3o Reserve price effect on revenue 3 a The relevant FOC r 7 X0 r 1 o In particular note that the optimal reserve price does not depend on the number of bidders Intuition the reserve price is useful only when there is one bidder with value above r a With F uniform on 01 and X0 0 then r so that L lir r i a Without reserve price the expected revenue with two bidders is and it goes up to 157 with the optimal reserve price a 25 increase in seller39s profit I April 21 2008 25 3o Risk Averse Bidders Assume now that each bidder has a utility function u R a R such that u0 0 u is strictly increasing and concave Proposition With symmetric independent private values the expected revenue in a first price auction is greater than that in a second price auction 0 First notice that the proof for the equilibrium strategy in a second price auction still applies with risk aversion gt still optimal for the risk averse bidder to bid its own value o In a first price auction consider 39y 07w gt R the equilibrium bidding strategy with 39y0 0 Assume that 39y is increasing and differentiable I ill April 21 2008 2 6 30 Risk Averse Bidders 2 o Nash equilibrium all other bidders are playing 39y It must be optimal for bidder 1 to bid 39yx o If bidder 1 bids 2 proba of winning is G y 1z and utility conditional on winning is uxi z o Equilibrium condition 39yx argmaxZ G 4 H X 7 z where G FA 1 0 Using FOC mgxux 7 39yx GXLlX7 39yx or 7 Xi YlXD V X mew ax I April 21 2008 21 3o Risk Averse Bidders 3 0 With risk neutrality B X X7 x 469 a Note that 39y0 O 30 0 Because u is strictly concave and u0 0 uy 2 u yy so that W 2 x7 x 0 Thus 3X 2 39yx implies 39y x 2 x This combined with 39y0 O 30 implies that 3X lt 39yx for X gt 0 gt Risk Aversion causes an increase in equilibrium bids Risk averse bidders wants to hedge themselves against the possibility of losing the auction I April 21 2008 28 3o
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'