Solutions Manual Foundations of Mathematical Economics Michael Carter November 15, 2002 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved Chapter 1: Sets and Spaces 1.1 {1,3,5,7...} or {πβπ :π is odd} 1.2 Every π₯ β π΄ also belongs to π΅. Every π₯ β π΅ also belongs to π΄. Hence π΄,π΅ have precisely the same elements. 1.3 Examples of ο¬nite sets are β the letters of the alphabet {A, B, C, ... , Z} β the set of consumers in an economy β the set of goods in an economy β the set of players in a game. Examples of inο¬nite sets are β the real numbers β β the natural numbers π β the set of all possible colors β the set of possible prices of copper on the world market β the set of possible temperatures of liquid water. 1.4 π ={1,2,3,4,5,6},πΈ ={2,4,6}. 1.5 The player set is π ={Jenny,Chris}. Their action spaces are π΄ ={Rock,Scissors,Paper} π=Jenny,Chris π 1.6 The setofplayersis π ={1,2,...,π}. The strategyspace ofeachplayeris the set of feasible outputs π΄ ={π ββ :π β€π } π π + π π where π is the output of dam π. π 1.7 The player set is π ={1,2,3}. There are 23 =8 coalitions, namely π«(π)={β ,{1},{2},{3},{1,2},{1,3},{2,3},{1,2,3}} There are 210 coalitions in a ten player game. 1.8 Assume that π₯ β (π βͺπ)π. That is π₯ β/ π βͺπ. This implies π₯ β/ π and π₯ β/ π, or π₯ β ππ and π₯ β ππ. Consequently, π₯ β ππ β©ππ. Conversely, assume π₯ β ππβ©ππ. This implies that π₯ β ππ and π₯ β ππ. Consequently π₯ β/ π and π₯ β/ π and therefore π₯β/ πβͺπ. This implies that π₯β(πβͺπ)π. The other identity is proved similarly. 1.9 βͺ π =π πβπ β© π =β πβπ 1 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved π₯ 2 1 π₯ 1 -1 0 1 -1 Figure 1.1: The relation {(π₯,π¦):π₯2+π¦2 =1} 1.10 The sample space of a single coin toss is {π»,π}. The set of possible outcomes in three tosses is the product { {π»,π}Γ{π»,π}Γ{π»,π}= (π»,π»,π»),(π»,π»,π),(π»,π,π»), } (π»,π,π),(π,π»,π»),(π,π»,π),(π,π,π»),(π,π,π) A typical outcome is the sequence (π»,π»,π) of two heads followed by a tail. 1.11 π β©βπ ={0} + where0=(0,0,...,0)istheproductionplanusingnoinputsandproducingnooutputs. To see this, ο¬rst note that 0 is a feasible production plan. Therefore, 0 β π. Also, 0ββπ and therefore 0βπ β©βπ. + + Toshowthatthereisnootherfeasibleproductionplaninβπ, weassumethecontrary. + That is, we assume there is some feasible production plan yββπ β{0}. This implies + the existence of a plan producing a positive output with no inputs. This technological infeasible, so that π¦ β/ π. 1.12 1. Let x β π(π¦). This implies that (π¦,βx) βπ. Let xβ² β₯ x. Then (π¦,βxβ²)β€ (π¦,βx) and free disposability implies that (π¦,βxβ²)βπ. Therefore xβ² βπ(π¦). 2. Again assume x β π(π¦). This implies that (π¦,βx) β π. By free disposal, (π¦β²,βx)βπ for every π¦β² β€π¦, which implies that xβπ(π¦β²). π(π¦β²)βπ(π¦). 1.13 The domain of β<β is {1,2}=π and the range is {2,3}β«π. 1.14 Figure 1.1. 1.15 Therelationβisstrictlyhigherthanβistransitive,antisymmetricandasymmetric. It is not complete, reο¬exive or symmetric. 2 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved 1.16 The following table lists their respective properties. < β€ = β β reο¬exive Γ β β β transitive β β symmetric Γ β asymmetric Γ Γ β β β anti-symmetric β β complete Γ Note that the properties of symmetry and anti-symmetry are not mutually exclusive. 1.17 LetβΌbeanequivalencerelationofasetπ β=β . Thatis,therelationβΌisreο¬exive, symmetricandtransitive. We ο¬rstshowthateveryπ₯βπ belongsto someequivalence class. Let π be any element in π and let βΌ (π) be the class of elements equivalent to π, that is βΌ(π)β‘{π₯βπ :π₯βΌπ} Since βΌ is reο¬exive, πβΌπ and so πββΌ(π). Every πβπ belongs to some equivalence class and therefore βͺ π = βΌ(π) πβπ Next, we show that the equivalence classes are either disjoint or identical, that is βΌ(π)β=βΌ(π) if and only if fβΌ(π)β©βΌ(π)=β . First, assume βΌ(π)β©βΌ(π)=β . Then πββΌ(π) but πβ/ βΌ(π). Therefore βΌ(π)β=βΌ(π). Conversely, assume βΌ(π)β©βΌ(π) β= β and let π₯ β βΌ(π)β©βΌ(π). Then π₯ βΌ π and by symmetry π βΌ π₯. Also π₯ βΌ π and so by transitivity π βΌ π. Let π¦ be any element in βΌ(π) so that π¦ βΌ π. Again by transitivity π¦ βΌ π and therefore π¦ β βΌ(π). Hence βΌ(π)ββΌ(π). Similar reasoning implies that βΌ(π)ββΌ(π). Therefore βΌ(π)=βΌ(π). We conclude that the equivalence classes partition π. 1.18 Thesetofpropercoalitionsisnotapartitionofthesetofplayers,sinceanyplayer can belong to more than one coalition. For example, player1 belongs to the coalitions {1}, {1,2} and so on. 1.19 π₯β»π¦ =β π₯βΏπ¦ and π¦ ββΏπ₯ π¦ βΌπ§ =β π¦ βΏπ§ and π§ βΏπ¦ Transitivity of βΏ implies π₯βΏπ§. We need to show that π§ ββΏπ₯. Assume otherwise, that is assume π§ βΏ π₯ This implies π§ βΌ π₯ and by transitivity π¦ βΌ π₯. But this implies that π¦ βΏπ₯ whichcontradicts the assumption that π₯β»π¦. Thereforewe conclude that π§ ββΏπ₯ and therefore π₯β»π§. The other result is proved in similar fashion. 1.20 asymmetric Assume π₯β»π¦. π₯β»π¦ =β π¦ ββΏπ₯ while π¦ β»π₯ =β π¦ βΏπ₯ Therefore π₯β»π¦ =β π¦ ββ»π₯ 3 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved transitive Assume π₯β»π¦ and π¦ β»π§. π₯β»π¦ =β π₯βΏπ¦ and π¦ ββΏπ₯ π¦ β»π§ =β π¦ βΏπ§ and π§ ββΏπ¦ Since βΏ is transitive, we conclude that π₯βΏπ§. It remains to show that π§ ββΏ π₯. Assume otherwise, that is assume π§ βΏ π₯. We know that π₯βΏπ¦ and transitivity implies that π§ βΏπ¦, contrary to the assumption that π¦ β»π§. We conclude that π§ ββΏπ₯ and π₯βΏπ§ and π§ ββΏπ₯ =β π₯β»π§ This shows that β» is transitive. 1.21 reο¬exive Since βΏ is reο¬exive, π₯βΏπ₯ which implies π₯βΌπ₯. transitive Assume π₯βΌπ¦ and π¦ βΌπ§. Now π₯βΌπ¦ ββ π₯βΏπ¦ and π¦ βΏπ₯ π¦ βΌπ§ ββ π¦ βΏπ§ and π§ βΏπ¦ Transitivity of βΏ implies π₯βΏπ¦ and π¦ βΏπ§ =β π₯βΏπ§ π§ βΏπ¦ and π¦ βΏπ₯ =β π§ βΏπ₯ Combining π₯βΏπ§ and π§ βΏπ₯ =β π₯βΌπ§ symmetric π₯βΌπ¦ ββ π₯βΏπ¦ and π¦ βΏπ₯ ββ π¦ βΏπ₯ and π₯βΏπ¦ ββ π¦ βΌπ₯ 1.22 reο¬exive Every integer is a multiple of itself, that is π=1π. transitive Assume π=ππ and π=ππ where π,π βπ. Then π=πππ so that π is a multiple of π. not symmetric If π = ππ, π β π, then π = 1π and π β/ π. For example, 4 is a π multiple of 2 but 2 is not a multiple of 4. 1.23 [π,π]={π,π¦,π,π§} (π,π)={π¦} 1.24 βΏ(π¦)={π,π¦,π§} β»(π¦)={π,π§} βΎ(π¦)={π,π₯,π¦} βΊ(π¦)={π,π₯} 4 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved 1.25 Let π be ordered by βΏ. π₯ β π is a minimal element there is no element which strictly precedes it, that is there is no element π¦ β π such that π¦ βΊ π₯. π₯ β π is the ο¬rst element if it precedes every other element, that is π₯βΎπ¦ for all π¦ βπ. 1.26 The maximal elements of π are π and π§. The minimal element of π is π₯. These are also best and worst elements respectively. 1.27 Assume that π₯ is a best element in π orderedby βΏ. That is, π₯βΏπ¦ for all π¦ βπ. Thisimpliesthatthereisnoπ¦ βπ whichstrictlydominatesπ₯. Therefore,π₯ismaximal in π. In Example 1.23, the numbers 5,6,7,8,9 are all maximal elements, but none of them is a best element. 1.28 Assumethattheelementsaredenotedπ₯ ,π₯ ,...,π₯ . Wecanidentifythemaximal 1 2 π element by constructing another list using the following recursive algorithm π =π₯ 1 1 { π₯ if π₯ β»π π = π π πβ1 π π otherwise πβ1 By construction, there is no π₯ which strictly succedes π . π is a maximal element. π π π 1.29 π₯β is maximal ββ there does not exist π₯β»π₯β that is β»(π₯β)={π₯:π₯β»π₯β}=β π₯β is best ββ π₯β βΏπ₯ for every π₯βπ ββ π₯βΎπ₯β for every π₯βπ That is, every π₯βπ belongs to βΎ(π₯β) or βΎ(π₯β)=π. 1.30 Let π΄ be a nonempty set of a set π ordered by βΏ. π₯ β π is a lower bound for π΄ if it precedes every element in π΄, that is π₯ βΎ π for all π β π΄. It is a greatest lower bound if it dominates every lower bound, that is π₯βΏπ¦ for every lower bound π¦ of π΄. 1.31 Any multiple of 60 is an upper bound for π΄. Thus, the set of upper bounds of π΄ is {60,120,240,...}. The least upper bound of π΄ is 60. The only lower bound is 1, hence it is the greatest lower bound. 1.32 The least upper bounds of interval [π,π] are π and π§. The least upper bound of (π,π) is π¦. 1.33 π₯ is an upper bound of π΄ ββ π₯βΏπ for every πβπ΄ ββ πβΎπ₯ for every πβπ΄ ββ π΄ββΎ(π₯) Similarly π₯ is a lower bound of π΄ ββ π₯βΎπ for every πβπ΄ ββ πβΏπ₯ for every πβπ΄ ββ π΄ββΏ(π₯) 5 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved 1.34 For every π₯ββ2, π₯β»π¦ if π₯ >π¦ or π₯ =π¦ and π₯ >π¦ 1 1 1 1 2 2 Since all elements π₯ββ2 are comparable, β» is complete; it is a total order. 1.35 Assume βΏ is complete for every π. Then for every π₯,π¦ β π and for all π = π 1,2,...,π, either π₯ βΏ π¦ or π¦ βΏ π₯ or both. Either π π π π π π π₯ βΌ π¦ for all π Then deο¬ne π₯βΌπ¦. π π π π₯ ββΌ π¦ for some π Let π be the ο¬rst individual with a strict preference, that is π = π π π min (π₯ ββΌπ¦ ). (Completeness of βΏ ensures that π is deο¬ned). Then deο¬ne π π π π π₯β»π¦ if π₯ β» π¦ π π π π¦ β»π₯ otherwise 1.36 Let π, π and π be subsets of a ο¬nite set π. Set inclusion β is reο¬exive since π βπ. transitive since π βπ and π βπ implies π βπ. anti-symmetric since π βπ and π βπ implies π =π Therefore β is a partial order. 1.37 Assume π₯ and π¦ are both least upper bounds of π΄. That is π₯ βΏ π for all π β π΄ and π¦ βΏ π for all π β π΄. Further, if π₯ is a least upper bound, π¦ βΏ π₯. If π¦ is a least upper bound, π₯βΏπ¦. By anti-symmetry, π₯=π¦. 1.38 π₯βΌπ¦ =β π₯βΏπ¦ and π¦ βΏπ₯ which implies that π₯=π¦ by antisymmetry. Each equivalence class βΌ(π₯)={π¦ βπ :π¦ βΌπ₯} comprises just a single element π₯. 1.39 maxπ«(π)=π and minπ«(π)=β . 1.40 The subset {2,4,8} forms a chain. More generally, the set of integer powers of a given number {π,π2,π3,...} forms a chain. 1.41 Assume π₯ and π¦ are maximal elements of the chain π΄. Then π₯ βΏ π for all π β π΄ and in particular π₯ βΏ π¦. Similarly, π¦ βΏ π for all π β π΄ and in particular π¦ βΏ π₯. Since βΏ is anti-symmetric, π₯=π¦. 1.42 1. By assumption, foreveryπ‘βπβπ,βΊ(π‘)is anonempty ο¬nite chain. Hence, it has a unique maximal element, π(π‘). 2. Let π‘ be any node. Either π‘ is an initial node or π‘ has a unique predecessor π(π‘). Either π(π‘) is an initial node, or it has a unique predecessor π(π(π‘)). Continuing in this way, we trace out a unique path from π‘ back to an initial node. We can be sure of eventually reaching an initial node since π is ο¬nite. 1.43 (1,2)β¨(3,1)=(3,2) and (1,2)β§(3,2)=(1,2) 6 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved 1.44 1. π₯β¨π¦ isanupperboundfor{π₯,π¦},thatisxβ¨y βΏπ₯andxβ¨yβΏπ¦. Similarly, π₯β¨π¦ is a lower bound for {π₯,π¦}. 2. Assume π₯ βΏ π¦. Then π₯ is an upper bound for {π₯,π¦}, that is π₯ βΏ π₯β¨π¦. If π is any upper bound for {π₯,π¦}, then π βΏ π₯. Therefore, π₯ is the least upper bound for {π₯,π¦}. Similarly, π¦ is a lower bound for {π₯,π¦}, and is greater than any other lower bound. Conversely,assume π₯β¨π¦ =π₯. Then π₯ is an upper bound for {π₯,π¦}, that is π₯βΏπ¦. 3. Using the preceding equivalence π₯βΏπ₯β§π¦ =β π₯β¨(π₯β§π¦)=π₯ π₯β¨π¦ βΏπ₯ =β (π₯β¨π¦)β§π₯=π₯ 1.45 A chain π is a complete partially ordered set. For every π₯,π¦ β π with π₯ β= π¦, either π₯β»π¦ or π¦ β»π₯. Therefore, deο¬ne the meet and join by { π¦ if π₯β»π¦ π₯β§π¦ = π₯ if π¦ β»π₯ { π₯ if π₯β»π¦ π₯β¨π¦ = π¦ if π¦ β»π₯ π is a lattice with these operations. 1.46 Assumeπ andπ arelattices, andletπ =π Γπ . Consideranytwoelements 1 2 1 2 x = (π₯ ,π₯ ) and y = (π¦ ,π¦ ) in π. Since π and π are lattices, π = π₯ β¨π¦ β π 1 2 1 2 1 2 1 1 1 1 and π = π₯ β¨π¦ β π , so that b = (π ,π ) = (π₯ β¨π¦ ,π₯ β¨π¦ ) β π. Furthermore 2 2 2 2 1 2 1 1 2 2 b βΏ x and b βΏ y in the natural product order, so that b is an upper bound for the {x,y}. Every upper bound bΛ = (Λπ ,Λπ ) of {x,y} must have π βΏ π₯ and π βΏ π¦ , 1 2 π π π π π π so that bΛ βΏ b. Therefore, b is the least upper bound of {x,y}, that is b = xβ¨y. Similarly, xβ§y=(π₯ β§π¦ ,π₯ β§π¦ ). 1 1 2 2 1.47 Let π be a subset of π and let πβ ={π₯βπ :π₯βΏπ for every π βπ} be the set of upper bounds ofπ. Then π₯β βπβ β=β . By assumption, πβ has a greatest lower bound π. Since every π β π is a lower bound of πβ, π βΏ π for every π β π. Therefore π is an upper bound of π. Furthermore, π is the least upper bound of π, since πβΎπ₯ for every π₯βπβ. This establishes that every subset of π also has a least upper bound. In particular, every pair of elements has a least upper and a greatest lower bound. Consequently π is a complete lattice. 1.48 Without loss of generality, we will prove the closed interval case. Let [π,π] be an interval in a lattice πΏ. Recall that π = inf[π,π] and π = sup[π,π]. Choose any π₯,π¦ in [π,π]βπΏ. Since πΏ is a lattice, π₯β¨π¦ βπΏ and π₯β¨π¦ =sup{π₯,π¦}βΎπ Therefore π₯β¨π¦ β [π,π]. Similarly, π₯β§π¦ β [π,π]. [π,π] is a lattice. Similarly, for any subsetπ β[π,π]βπΏ,supπ βπΏifπΏiscomplete. Also,supπ βΎπ=sup[π,π]. Therefore supπ β[π,π]. Similarly infπ β[π,π] so that [π,π] is complete. 7 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved 1.49 1. The strong set order βΏ is π antisymmetric Let π ,π β π with π βΏ π and π βΏ π . Choose π₯ β π 1 2 1 π 2 2 π 1 1 1 and π₯ βπ . Since π βΏ π , π₯ β¨π₯ βπ and π₯ β§π₯ βπ . On the other 2 2 1 π 2 1 2 1 1 2 2 hand, since π βΏπ , π₯ =(π₯ β¨(π₯ β§π₯ )βπ andπ₯ =π₯ β§(π₯ β¨π₯ )βπ 2 1 1 1 1 2 2 2 2 1 2 1 (Exercise 1.44. Therefore π =π and βΏ is antisymmetric. 1 2 π transitive Let π ,π ,π β π with π βΏ π and π βΏ π . Choose π₯ β π , 1 2 3 1 π 2 2 π 3 1 1 π₯ β π and π₯ β π . Since π βΏ π and π βΏ π , π₯ β¨π₯ and π₯ β§π₯ 2 2 3 3 1 π 2 2 π 3 1 2 2 3 are in π . Therefore π¦ =π₯ β¨(π₯ β§π₯ )βπ which implies 2 2 1 2 3 2 ( ) π₯ β¨π₯ =π₯ β¨ (π₯ β§π₯ )β¨π₯ 1 3 1 2 3 3 ( ) = π₯ β¨(π₯ β§π₯ ) β¨π₯ 1 2 3 3 =π¦ β¨π₯ βπ 2 3 3 since π βΏ π . Similarly π§ =(π₯ β¨π₯ )β§π₯ βπ and 2 π 3 2 1 2 3 2 ( ) π₯ β§π₯ = π₯ β§(π₯ β¨π₯ ) β§π₯ 1 3 1 1 2 3 ( ) =π₯ β§ (π₯ β¨π₯ )β§π₯ 1 1 2 3 =π₯ β§π§ βπ 1 2 1 Therefore, π βΏ π . 1 π 3 2. π βΏ π if and only if, for every π₯ ,π₯ β π, π₯ β¨π₯ β π and π₯ β§π₯ β π, which π 1 2 1 2 1 2 is the case if and only if π is a sublattice. 3. Let πΏ(π) denote the set of all sublattices of π. We have shown that βΏ is π reο¬exive, transitive and antisymmetric on πΏ(π). Hence, it is a partial order on πΏ(π). 1.50 Assume π βΏ π . For any π₯ βπ and π₯ β π , π₯ β¨π₯ β π and π₯ β§π₯ β π . 1 π 2 1 1 2 2 1 2 1 1 2 2 Therefore supπ βΏπ₯ β¨π₯ βΏπ₯ for every π₯ βπ 1 1 2 2 2 2 which implies that supπ βΏsupπ . Similarly 1 2 infπ βΎπ₯ β§π₯ βΎπ₯ for every π₯ βπ 2 1 2 1 1 1 which implies that infπ βΎ infπ . Note that completeness ensures the existence of 2 1 supπ and infπ respectively. 1.51 An argument analogous to the preceding exercise establishes =β . (Complete- ness is not required, since for any interval π=inf[π,π] and π=sup[π,π]). To establish the converse, assume that π = [π ,π ] and π = [π ,π ]. Consider any 1 1 1 2 2 2 π₯ βπ and π₯ βπ . There are two cases. 1 1 2 2 Case 1. π₯ βΏπ₯ Since π is a chain, π₯ β¨π₯ =π₯ βπ . π₯ β§π₯ =π₯ βπ . 1 2 1 2 1 1 1 2 2 2 Case 2. π₯ βΊπ₯ Since π is a chain, π₯ β¨π₯ = π₯ . Now π βΎ π₯ βΊ π₯ βΎ π βΎ π . 1 2 1 2 2 1 1 2 2 2 Therefore, π₯ = π₯ β¨π₯ β π . Similarly π βΎ π βΎ π₯ βΊ π₯ βΎ π . Therefore 2 1 2 1 2 1 1 2 2 π₯ β§π₯ =π₯ βπ . 1 2 1 2 We have shown that π βΏ π in both cases. 1 π 2 1.52 Assume that βΏ is a complete relation on π. This means that for every π₯,π¦ βπ, either π₯βΏπ¦ or π¦ βΏπ₯. In particular, letting π₯=π¦, π₯βΏπ₯ for π₯βπ. βΏ is reο¬exive. 8 βc 2001 Michael Carter Solutions for Foundations of Mathematical Economics All rights reserved 1.53 Anti-symmetryimpliesthateachindiο¬erenceclasscontainsasingleelement. Ifthe consumerβspreferencerelationwasanti-symmetric,therewouldbe nobasketsofgoods between which the consumer was indiο¬erent. Each indiο¬erence curve which consist a single point. 1.54 We previously showed (Exercise 1.27) that every best element is maximal. To provethe converse,assumethatπ₯ismaximalinthe weaklyorderedsetπ. Wehaveto show that π₯βΏπ¦ for all π¦ βπ. Assume otherwise, that is assume there is some π¦ βπ for which π₯ ββΏ π¦. Since βΏ is complete, this implies that π¦ β» π₯ which contradicts the assumption that π₯ is maximal. Hence we conclude that π₯ βΏ π¦ for π¦ β π and π₯ is a best element. 1.55 False. Achainhasatmostonemaximalelement(Exercise1.41). Here,uniqueness is ensured by anti-symmetry. A weakly ordered set in which the order is not anti- symmetric may have multiple maximal and best elements. For example, π and π are both best elements in the weakly ordered set {πβΌπβ»π}. 1.56 1. For every π₯βπ, either π₯βΏ π¦ =β π₯ββΏ(π¦) or π¦ βΏπ₯ =β π₯ββΎ(π¦) since βΏ is complete. Consequently, βΏ(π¦)βͺβΊ(π¦) = π If π₯ β βΏ(π¦)β©βΎ(π¦), then π₯ βΏ π¦ and π¦ βΏπ₯ so that π₯βΌπ¦ and π₯βπΌ . π¦ 2. For every π₯βπ, either π₯βΏπ¦ =β π₯ββΏ(π¦) or π¦ β»π₯ =β π₯ββΊ(π¦) since βΏ is complete. Consequently, βΏ(π¦)βͺβΊ(π¦)=π and βΏ(π¦)β©βΊ(π¦)=β . 3. For every π¦ β π, β»(π¦) and πΌ partition βΏ(π¦) and therefore β»(π¦), πΌ and βΊ(π¦) π¦ π¦ partition π. 1.57 Assume π₯βΏπ¦ andπ§ ββΏ(π₯). Then π§ βΏπ₯βΏπ¦ by transitivity. Thereforeπ§ ββΏ(π¦). This shows that βΏ(π₯)ββΏ(π¦). Similarly, assume π₯ β» π¦ and π§ β β»(π₯). Then π§ β» π₯ β» π¦ by transitivity. Therefore π§ β β»(π¦). This shows that βΏ(π₯) β βΏ(π¦). To show that βΏ(π₯) β= βΏ(π¦), observe that π₯ββ»(π¦) but that π₯β/ β»(π₯) 1.58 Every ο¬nite ordered set has a least one maximal element (Exercise 1.28). 1.59 Kreps(1990,p.323),Luenberger(1995,p.170)andMas-Colelletal. (1995,p.313) adopt the weak Pareto order, whereas Varian (1992, p.323) distinguishes the two or- ders. Osborneand Rubinstein (1994,p.7) also distinguish the two orders, utilizing the weak order in deο¬ning the core (Chapter 13) but the strong Pareto order in the Nash bargaining solution (Chapter 15). 1.60 Assumethatagroupπ isdecisiveoverπ₯,π¦ βπ. Letπ,πβπ betwootherstates. We have to show that π is decisive over π and π. Without loss of generality, assume for all individuals π βΏ π₯ and π¦ βΏ π. Then, the Pareto order implies that π β» π₯ and π π π¦ β»π. Assume that for every π β π, π₯ βΏ π¦. Since π is decisive over π₯ and π¦, the social π order ranks π₯βΏπ¦. By transitivity, πβΏπ. By IIA, this holds irrespective of individual preferences on other alternatives. Hence, π is decisive over π and π. 1.61 Assume that π is decisive. Let π₯, π¦ and π§ be any three alternatives and assume π₯βΏπ¦ for every πβπ. Partition π into two subgroups π and π so that 1 2 π₯βΏ π§ for every πβπ and π§ βΏ π¦ for every πβπ π 1 π 2 Since π is decisive, π₯βΏπ¦. By completeness, either π₯βΏπ§inwhichcaseπ isdecisiveoverπ₯andπ§. Bytheο¬eldexpansionlemma(Exercise 1 1.60), π is decisive. 1 9