Economics 201b Spring 2010 Problem Set 6 Solution

1. There are many foreign embassies in the Washington, DC. In fact, there is an area of the city where quite a few of them are very close to each other. As you walk along one of the streets you observe three embassies standing next to each other. Each embassy has a agpole with its national ag ying in the wind. You know that the agpole height chosen by each embassy depends continuously on the heights chosen by other two embassies. (For instance, having too tall a pole compared to the neighbors would be ostentatious, whereas having one too short would look stingy.) Moreover, having observed heights chosen by the neighbors, each embassy has a single, most favorite height to set. DC ordinance imposes an upper limit of 100 feet on agpole heights of the embassies. The choices of agpole heights are in equilibrium when no one wishes to change the height of their agpole. Prove that there exists an equilibrium. Solution. To begin, note how little is known about the optimal height choice by embassies. It might be the case that neither of embassies would like the have the highest or the lowest agpole, then it is clear that there is an equilibrium when all agpole heights converge to some average value. However, it is also (i)+100 , i.e. always possible that some embassy would like to set its height as maxi h2 to have (one of the) highest agpole(s). All we know is that embassy choice is a continuous function on a closed and bounded set in R3 + . But, this information is sucient to apply a xed-point argument. Notice that is exactly how we prove the existence of equilibrium in AD economy. So, dene closed and bounded set D = [0, 100] [0, 100] [0, 100] R3 + . Heine, moreover, it is clear that it is Borrel theorem implies that D is compact in R3 + convex. Now, dene f : D D as a function that encodes the optimal choice of each embassy, given the choices of its neighbors. Brouwers x point theorem suggest that x D f (x ) = x , i.e. there is an equilibrium when no one wishes to change the height of their agpole. 2. Consider a two-person, two-good exchange economy where all agents have the same utility function, i = 1, 2: u(x1i , x2i ) = max{2 min{2x1i , x2i }, min{x1i , 4x2i }}, 1 = 2 = (1, 1).

(a) Draw the indierence curves for one of the consumer. Are this consumers preferences convex? Solution. The preferences are not convex: the indierence curves have x and x2 = 1 x . See Figure 1. kinks along the lines x2 = 2x1 , x2 = 1 2 1 4 1

Economics 201b Spring 2010

Problem Set 6 Solution

(b) Draw the Edgeworth box for this economy, denoting Pareto set, individually rational and core allocations. Solution. We know that in Edgeworth box economy, contract curve is equal to the core of the economy, thus, it is a set of all individually rational Pareto optimal allocations. See Figure 2. (c) Now suppose we have an economy of I N identical consumers with I 2, each of which has the same preferences as the consumers described above and endowments are (1 , 2 ) = (1 + 3, 2 ) where (0, 1). Find a necessary and sucient condition on that must be satised for there to exist a Walrasian equilibrium of this economy. Show that as I increases, the set of a (0, 1) that satisfy the condition you found increases in size. Solution. As with any competitive equilibrium problem, it helps to rst characterize each agents demand as a function of prices. Here, this can be summed up by considering three cases for the price vector: 1 p1 > , p2 3 p1 1 If = , p2 3 p1 1 If < , p2 3

If

then x2 (p1 , p2 ) = 2x1 (p1 , p2 ). then x2 (p1 , p2 ) = 2x1 (p1 , p2 ) 1 x1 (p1 , p2 ). or x2 (p1 , p2 ) = 4 1 then x2 (p1 , p2 ) = 2x1 (p1 , p2 ). 4

All consumers in this economy are identical, so if our equilibrium prices p1 > 1 , then we have x are characterized by p 2 = 2x1 , and market clearing 3 2 implies that we must have 2 if i = 21i i. However, this is only possible p 1 = 0, and we have constrained (0, 1). Thus, we cannot have p > 1 3 2 in equilibrium. p 1 < 1 as this would require = 1. Thus, Similarly, we also cannot have p 3
2 1 in any equilibrium, we must have p = 1 . In this case, some agents can 3 2 consume on the x2 = 2x1 line, while others consume on the x2 = 1 x line 4 1 (and note that the endowment will lie somewhere between these two lines, depending on the value of ). Now we are in a position to show when equilibria exist. Denote a type a agent as one who consumes on x2 = 2x1 line and a type b agent 1 as one who consumes on x2 = 4 x1 line, and further let Ia and Ib Z+ denote the number of agents of each type in equilibrium. In equilibrium we must satisfy three equations: budget balancing for the a agents, budget balancing for the b agents, and market clearing for one of the goods (well use good 1). We set p1 = 1 and p2 = 3. The wealth of each type of agent is Wa = Wb = 1 + 3 + 3(2 ) = 7, which is independent of . We may now write the

Economics 201b Spring 2010

Problem Set 6 Solution

budget constraint for the a agents as x1a + 3x2a = 7. Applying x2a = 2x1a yields that x1a = 1 and x2a = 2 in an equilibrium. Similarly, we have 4x2b = x1b , implying that x1b = 4 and x2b = 1. Market clearing for good 1 implies that Ia x1a + Ib x1b = (Ia + Ib )(1 + 3) (i.e. the total demand for good 1 equals the total endowment of good 1). Solving this for yields: = Ib I

That is, must be equal to the ratio of type b agents to the total number of agents. What values could this ratio take? For two agents, this ratio , or 1. However, we cannot have = 0 or = 1, so could be equal to 0, 1 2 the only option = 1 , which yields an equilibrium with one agent of each 2 type. Similarly, for three total agents, we can have = 1 or = 2 . That is, 3 3 1 for = 3 , in equilibrium the agents will sort themselves so that two are of type a while one is of type b, while for = 2 there will be one agent 3 of type a and two of type b (recall that the agents are indierent between p 1 1 =3 ). either type at p
2

Generalizing this, our necessary and sucient condition on , given the total number of agents I , is: A= n : n N and n < I I

Note that as I increases, the number of elements in the set A increases as well: each one-agent increase in I increases the set of possible values for by one. (d) Explain in a few sentences how these results relate to Theorem 2 in the Lecture Notes 11. That is, relate your above results to the fact that, in this economy, we can show that (0, 1) p 0, p 0 with 0 conE (p ) and xi Di (p ) such that 1 I
L I I

p
=1

i=1

x i

i=1

i
l

2L max{ i I

: i = 1, ..., I }

Fix = 1 and compute an explicit bound for the market value of the 3 surpluses and shortages in the economy. Verify that the bound provided by the theorem is tight enough. Solution. We have just showed above that more and more values of can generate economies with exact Walrasian equilibria, when we add more agents to this economy (i.e., when we increase I ). Thus, when we take any (0, 1), the greater the number of agents, the more likely it is that the we pick will be close to one of the the values of that generates an exact equilibrium. This means that the excess demands given will not be too large for large I , which is what is predicted by Theorem 2. 3

Economics 201b Spring 2010

Problem Set 6 Solution

Lets compute the explicit bound for the market value of the surpluses and shortages in the economy and verify that the bound provided by the theorem is tight enough. In our case, there are two type of agents, Ia and Ib , with identical endowments. Consequently, we need to show that (0, 1) p 0, p 0 with 0 conE (p ) and x i Di (p ) such that
p 1 |Ia x1a + (I Ia ) x1b I (1 + 3)| + + p 2 |Ia x2a + (I Ia ) x2b I (2 )| 4 max{1 + 3, 2 } 1 3 So, pick p = 4 , 4 0 . a-type agents demand is given by x a = (2, 1) and b-type agents demand is given by xb = (1, 4). We have

1 |Ia + 4(I Ia ) I (1 + 3)| + 4 3 + |2Ia + (I Ia ) I (2 )| 4 max{1 + 3, 2 } 4

Consider 3 following cases: I = 0 mod 3. In this just pick Ia = 2 I, 1 Ia I 1. We have |2I 3 3Ia | = 0 16. I = 1 mod 3. In this case I = h + 1, where h = 0 mod 3 and h 3. 2 Again, pick Ia = 3 h, 1 Ia I 1. We have |2I 3Ia | = |2h + 2 2h| = 2 16. I = 2 mod 3. Consider two possibilities, either I = 2, then Ia = 1 and |2I 3Ia | = 1 16, or I 5 and I = h + 2, where h = 0 mod 3 and h, 1 Ia I 1. We have |2I 3Ia | = h 3. Again, pick Ia = 2 3 |2h + 4 2h| = 4 16. So, we have veried that the bound provided by the theorem is tight enough. 3. Consider four-person, two-good pure exchange economy where agents have endowments 1 = 2 = (10, 10) and 3 = 4 = (10, 30) and the same utility function Ui (x1i , x2i ) = log x1i + log x2i , i = 1, 2, . . . , 4. For each allocation vector given below show whether the it is Pareto optimal; is in the core (if not, provide a blocking coalition); can be supported as a competitive equilibria for some price vector. Explain your reasoning. (a) x1 = x2 = (7.5, 15) and x3 = x4 = (12.5, 25).

Economics 201b Spring 2010

Problem Set 6 Solution

Solution. The allocation x1 = x2 = (7.5, 15) and x3 = x4 = (12.5, 25) is Pareto optimal, is in the core and can be supported as a competitive equip libria for some price vector p1 = 2. To prove this claim, we will rst show 2 that this allocation is competitive equilibrium, thus, Pareto optimality follows from the First Welfare Theorem. Since any Walrasian equilibrium lies in the core by the Strong First Welfare Theorem, it has to be core allocation as well. Ui Ui |x1i =0+ = x |x2i =0+ = +, so consumer i will never deNotice that x 1i 2i mand zero amount of any commodity and the prices have to be strictly Ui \x1i p1 1 2i = p for all agents. Also, = x = p positive. Observe M RSi = U x1i 2 i \x2i since 80 221 + 223 2x21 + 2x23 1 1 = = = = p = . 40 211 + 213 2x11 + 2x13 p 2 Thus, an allocation x1 = x2 = (7.5, 15) and x3 = x4 = (12.5, 25) is indeed a Walrasian equilibrium with p = 1 . 2 (b) x1 = x2 = ( 50, 2 50) and x3 = x4 = (20 50, 40 2 50). Solution. The allocation x1 = x2 = ( 50, 2 50) and x3 = x4 = (20 50, 40 2 50) is a Pareto optimal, but it is not in the core, and, consequently, it cannot be Walrasian equilibrium. To see this, observe that the proposed allocation is a Pareto optimal one, since it is interior and MRS for all agents are equalized (since the utility functions are strongly monotone, any Pareto optimal allocation has to be interior). Now, notice that agents 1 and 2 receive the same utility in the proposed allocation as they would by just consuming their endowment. As it was shown in the lecture, such allocation cannot be in the core of 2fold replica economy, and, consequently, it cannot be supposed as competitive equilibrium. Similarly to what was done in lecture, consider a blocking coalition of agents 1, 2 and 3. It is easy to see that this coalition blocks the current allocation by dividing their aggregate endowment, for instance, in the following way: x1 = x2 = (8, 13) and x3 = (14, 24). (c) x1 = (8, 12), x2 = (9, 11), x3 = (12, 23) and x4 = (11, 29). Solution. The allocation x1 = (8, 12), x2 = (9, 11), x3 = (12, 23) and x4 = (11, 29) is neither Pareto optimal nor is in the core. It cannot be Pareto optimal because MRS are not equalized among agents, and (weak) Pareto optimality of core allocations implies that it cannot be in the core. Since all assumptions of the First Welfare Theorem are satised in this economy, the proposed allocation cannot be sustained as a Walrasian equilibrium because otherwise it has be Pareto optimal. 4. Give an example of a three-person, two-good pure exchange economy where all agents have the same utility function Ui (x1i , x2i ) = log x1i +log x2i ,. Find a set of 5

Economics 201b Spring 2010

Problem Set 6 Solution

integer endowments for these agents along with a Pareto optimal, individually rational integer allocation that is not in the core. Solution. Note that in any two-agent economy core is just a contract curve, i.e. set of all individually rational Pareto optimal allocations. So, to nd a counterexample to this fact, we need to consider three agents. Consider the endowments 1 = (1, 4), 2 = (4, 1), 3 = (1, 1) then x1 = x2 = x3 = (2, 2) is a Pareto optimal, individually rational allocation that is not in the core. Notice that the coalition of agents 1 and 2 can get together and split their goods down the middle so that each agent receives the preferred allocation (2.5, 2.5). But, it is clearly an individual rational one. So, we need only to verify Pareto optimality, and this is straightforward as well since the solution is in the interior, and utilities are smooth, it suces to check that the marginal rates of x2i for all agents and is 1 under substitution are equal. Since M RSi (x1i , x2i ) = x 1i the current allocation, we have Pareto optimality. 5. Consider a pure exchange economy with H = 2 consumers and L goods, with social endowment RL ++ . In this question, we will consider the n-fold replica of this economy. In the n-fold replica, there are 2n agents, of whom n (referred to as type 1 agents) have preferences and endowments identical to those of agent 1 in the original economy, and n (referred to as type 2 agents) have preferences and endowments identical to those of agent 2 in the original economy. (a) Let p be an equilibrium price vector for the original economy. Show that p is also an equilibrium price vector for the (larger) n-fold replica economy. Solution. Let p be an equilibrium price vector for the original economy. Then it must be the case that x 1 D (p , 1 ), x2 D (p , 2 ) and markets clear x 1 + x2 = 1 + 2 . Now, consider nfold replica economy. Since all type 1 and 2 agents have the same preference and endowments as in the original economy, their demand sets are unchanged when they face price p and we can show that markets clear in nfold replica economy. Indeed,
2n x i (p , i ) i=1 n 2n x i (p , i ) i=1 n (x 1 (p , 1 ) n 2n

= = =
i=1

x i (p , i ) =

i=n+1 x 2 (p , 2 )) 2n

= n(1 + 2 ) =

i +
i=n+1

i =
i=1

Thus, p is an equilibrium price vector in n-fold replica. (b) Now, lets consider a special case where there are two commodities and two type of agents. Type 1 is characterized as U1 (x11 , x21 ) = x11 x21 , 1 = (10, 0) 6

Economics 201b Spring 2010

Problem Set 6 Solution

and type 2 is characterized as U2 (x12 , x22 ) = (x12 ) 2 (x22 ) 2 , 2 = (0, 10). Show that the allocation x1 = x2 = (5, 5), (x1 to agents of type 1 and x2 to agents of type 2) is in the core for all levels of replication n. Solution. We will prove the claim by showing that the allocation x1 = x2 = (5, 5), is a (unique) competitive equilibrium of the original economy and then appeal to the Strong First Welfare Theorem to argue that it must be in the core for any level of replication n. Proving that x1 = x2 = (5, 5) is unique equilibrium is actually quite easy. We have symmetric CobbDouglas consumers, thus, everything is smooth, strictly quasi-concave and aggregate demand curves slope downwards. Consequently, we have a unique symmetric Walrasian equilibrium with p 1 = p2 and x1 = x2 = (5, 5).
1 1

(c) Continue to assume two-good, two-agent type economy given above. Show that the allocation x1 = (9, 9), x2 = (1, 1), is in the core for the original economy with one agent of the each type and is not in the core for the n-fold replica with n 2. Discuss. Solution. Notice that for the reason just given in (b) the set of all Pareto optimal allocations is just a diagonal of the Edgeworth box connecting two origins, O1 and O2 . Thus, an allocation x1 = (9, 9), x2 = (1, 1) is in the core because it satises individual rationality constraints, i.e. U1 (9, 9) > U1 (10, 0) and U2 (9, 9) > U2 (0, 10). To show that the allocation x1 = (9, 9), x2 = (1, 1) is not in the core for the n-fold replica with n 2, we will show that it is not in the core of 2-fold replica. Because any coalition that is active in n-fold replica is also active in (n+1)-replica, we obtain the result we seek. So, notice that current allocation achieves relatively low level of utility for type 2 agents. As it has been shown in the lecture, together with one type 1 agent they can block the allocation. For instance they block it with an allocation x1 = (6, 16) and x3 = x4 = (2, 2). 6. Give an example of acyclic preference relation that is not transitive. Solution. Note that any transitive preference relation is acyclic, this follows directly from denition of acyclicity. While we usually work with transitive preferences, sometimes it might be too strong of an assumption. As an example of acyclic preference relation that is not transitive consider following preference relation: an individual prefers apples (A) to bananas (B) and bananas to cherries (C). However, one is indierent between apples and cherries, or formally: A B C and A C 7

Economics 201b Spring 2010

Problem Set 6 Solution

This relation is clearly acyclic because there is no cycle (notice that would not be true if C A). But this preference relation is not quasi-transitive, and, thus, is not transitive, since quasi-transitivity would require A C.

Economics 201b Spring 2010

Problem Set 6 Solution

Economics 201b Spring 2010

Problem Set 6 Solution