Theory of General Economic Equilibrium Exercises to Chapter 4, page 1

Exercises to Chapter 4

(1) An allocation in the core must be Pareto optimal. If all bundles are in the interior of
R2+ , which they must be since they are individually rational, then marginal utilities of the
individual 1 and 3 must be equal,
3x11 1
= √ , (1)
2x12 2 x32
and the bundle of individual 2 must satisfy

x22 3
= , (2)
x21 2

and the bundles must satisfy the feasibility condition

x11 + x21 + x31 = 6

(3)
x12 + x22 + x32 = 8.

To this we first add the conditions of individual rationality,

2 √
u1 (x11 , x12 ) ≥ 4, u2 (x21 , x22 ) ≥ , u3 (x31 , x32 ) = 2 + 5, (4)
3
and for each coalition the condition that there is no improvement, that is
!
2
4, < int {(u1 (x10 ), u2 (x20 )) | x10 + x20 = (4, 3)}
3
√
!
2 (5)
, 2 + 5 < int {(u2 (x20 ), u3 (x30 )) | x20 + x30 = (4, 7)}
3
 √ 
4, 2 + 5 < int {(u1 (x10 ), u3 (x30 )) | x10 + x30 = (4, 6)}

Now, the core allocations are all ((x11 , x12 ), (x21 , x22 ), (x31 , x32 )) satisfying (1)-(5).

(2) A nonempty set N of allocations in E is a von Neumann-Morgenstern solution if

(i) each x = (x1 , . . . , xm ) is feasible, i.e. xi ∈ Xi , i =, 1, . . . , m, and mi=1 xi = mi=1 ωi .
P P

(ii) the set N is internally stable w.r.t. domination, i.e. there is no x0 ∈ N such that for some
S ⊆ N, i∈S xi0 = i∈S ωi and xi0 ∈ Pi (xi ), all i ∈ S ,
P P

(iii) the set N is externally stable w.r.t. domination, i.e. for each feasible allocation x < N,
there is an allocation xi00 ∈ N and a coalition T ⊆ N such that i∈T xi00 = i∈T ωi and
P P

xi00 ∈ Pi (xi ), all i ∈ T .

Theory of General Economic Equilibrium Exercises to Chapter 4, page 2

The core of an economy E consists of all the non-dominated solutions, so it is internally

stable, and it is also externally stable, since allocations not in the core must be dominated.
It follows that the economy has a nonempty von Neumann-Morgenstern solution if it has a
nonempty core, cf. Thm.4.2.

(3) For each k, an allocation xk in the k-replica economy Ek induces “equal-treatment” allo-
cations x s in all the economies Eks for s ∈ N, so that in this sense xk is an allocation in Eks
for all s. We show that if an allocation xk in Ek is in the core of Eks for all s ∈ N, then it is
(genuinely) equal-treatment. We shall need that preferences Pi for i = 1, . . . , m are complete
so that for each i and each pair (xi , xi0 ) of bundles in Xi , either xi0 ∈ cl Pi (xi ) or xi ∈ cl Pi (xi0 ).
Suppose that x ∈ Core(Ek ) but is not equal-treatment, so that there are at least two agents
i1 , i2 with identical characteristics such that xik1 , xik2 .
Assume that xik2 ∈ cl Pi (xik1 ). By strict convexity, we then have that

1 k 1 k
x∗ = x + x ∈ Pi (xik1 ).
2 i2 2 i1
In E2k , the allocation where each of the bundles xik for i = 1, . . . , m occurs twice can be
improved via the coalition consisting of the two copies of i1 together with two one of each
individual representing the bundles xik for i , i1 , i2 : If the two copies of i1 get xi and the other
participants keep their bundles, this allocation is feasible for the coalition, and upon some
redistributing it is possible to assign preferred bundles to all members of the coalition. It
follows that the ”equal-treatment” version of xk does not belong to Core(E2k ).

(4) A Walras equilibrium in the coalition production economy E = ((Xi , Pi , ωi )mi=1 , Y) is an

array (x, y, p, π), where x = (x1 , . . . , xm ) ∈ mi=1 Xi is an array of indvidually feasible bundles,
Q

y ∈ Y(N) a production vector feasible for the grand coalition, p ∈ 4 a price system, and
π = (π1 , . . . , πm ) ∈ Rm+ an array of profit shares, which satisfy
(i) aggregate feasibility, i.e. x mi=1 xi = mi=1 ωi + y,
P P

(ii) Individually optimal given p and π: for each i, p · xi ≤ p · ωi + πi , and if xi0 ∈ Pi (xi ),
then p · xi0 > p · ωi + πi , and p · y = maxy0 ∈Y(N) p · y0 },
(iii) no coalitional improvement: there is no S ⊆ N and yS ∈ Y(S ), (πSi )i∈S ∈ RS+ such that
i∈S πi = p · y and p · xi ≤ p · ωi + πi for some xi ∈ Pi (xi ), all i ∈ S .
P S S 00 S 00

For the second part, we first notice that if (x, y, p, π) is a Walras equilibrium, then π
belongs to the core of the TU game (N, v) with

v(S ) = max p · y.
y∈Y(S )

Indeed, if there is S with v(S ) > i∈S πi , then there is yS ∈ Y(S ) and (πSi )i∈S with i∈S πSi =
P P

p · yS and πSi > πi for each i ∈ S , and we get a contradiction of (iii).

Theory of General Economic Equilibrium Exercises to Chapter 4, page 3

If (x, y, p, π) is a Walras equilibrium and x does not belong to the core of E, then there
must be S ⊆ N and (xi0 )i∈S such that mi=1 (xi − ωi ) ∈ Y(S ) and xi0 ∈ Pi (xi ) for all i ∈ S . By the
P

properties of a Walras equililbrium, we have that

p · (xi0 − ωi ) > πi , all i ∈ S ,

and it follows that X X

v(S ) ≥ p · (xi0 − ωi ) > πi
i∈S i∈S

contradicting that π belongs to the core of (N, v).

(5) Consider an economy with 2 goods and 3 consumers, all having consumption set R2+ and
endowment (2, 2). The preferences can be described by utility functions which are positively
homogeneous (of first degree) and the set of bundles having utility 1 is defined as
(i) Consumer 1: The line segment from (2, 2) to (1, 3) together with the vertical half-line
from (1, 3) and the horizontal half-line from (2, 2),
(ii) Consumer 2: The line segment from (2, 2) to (3, 1) together with the vertical half-line
from (2, 2) and the horizontal half-line from (3, 1),
(iii) Consumer 3: The set {(x1 , x2 ) ∈ R2+ | x1 x2 = 1}.
Then the allocation ((1, 3), (3, 1), (2, 2)) defines a Walras equilibrium together with the price
vector (1, 1), consequently the allocation belongs to the core. All the marginal contributions
of the consumers are 4, so the allocation is a value allocation as well.

(6) The economy R3+ , P, ωi )3i=1 considered

 inBox 4 satisfies the Assumptions 0.1 and 0.2. As
shown in Box 4, the allocation x = 3 , 3 , 3 ∈ F (E) belongs to the bargaining set of E but
1 1 1

is not in the core, indeed the coalition {1, 2} has an improvement of x. But then the coalition
consisting of k copies of 1 and of 2 has an improvement of the equal-treatment allocation xk
in Ek induced by k, consequently xk is not in the core of Ek .