5.
The Welfare Theorems
1 The First Welfare Theorem
If preferences are locally nonsatiated and if (x∗ , y ∗ , p) is a price equilibrium with
transfers, then (x∗ , y ∗ ) is Pareto optimal. Specifically, any Walrasian equilib-
rium is Pareto optimal.
Vector (x∗ , y ∗ , p) is a price equilibrium
P with transfers P if there is an assign-
ment of wealth levels (ω1 , ..., ωI ) with i ωi = p · ω + j p · yj∗ such that:
1. Firms solve their PMPs
∗
2. For
P every
∗
P ∗ xi is maximal for <i in {xi ∈ Xi : p · xi ≤ ωi }
consumer,
3. i xi = ω + j yj
The sketch of the proof is that any allocation that Pareto dominates (x∗ , y ∗ )
must be not feasible. Because of local nonsatiation, p · xi ≥ ωi ∀i and p · xi > ωi
for some i. But then, the alternative allocation (x, y) is not feasible.
2 The Second Welfare Theorem
I J
Given an economy {(Xi , &i )}i=1 , {Yj }j=1 , ω with every Yj and &i convex
and locally nonsatiated, then for every Pareto optimal allocation (x∗ , y ∗ ) with
x∗ being such that every agent consumes a positive amount of each good, there
is a price vector p 6= 0 such that (x∗ , y ∗ , p) is a Walrasian equilibrium.
It is a partial reverse of the First Welfare Theorem regarding that it gives
conditions under which a Pareto Optimal allocation can be supported as a price
equilibrium with transfers. For instance, if we want to implement (x∗ , y ∗ ) as
a price equilibrium with transfers, we find the relevant prices p and give each
consumer income p · x∗i . That Pareto optimal allocation and the price p will be
a price equilibrium with transfers.
We can see this as a redistribution of initial endowments and profit shares
so that, under the relevant prices, (x∗ , y ∗ ) is an equilibrium, or as a lump sum
redistribution of wealth.
However, the informational requirements to implement the Pareto optimal
allocation (x∗ , y ∗ ) are too demanding, because preferences and endowments for
each individual have to be known. Furthermore, market completeness, price-
taking behavior and enforceability of the planner’s decision have to be assumed.
For instance, the presence of externalities (as we will discuss in Chapter 6)
usually leads to a situation where a Walrasian equilibrium is not Pareto optimal,
and thus the First Welfare Theorem does not hold.
3 Examples
We will finally study extreme situations that will help us to understand the
concepts of equilibrium and Pareto optimality.
1
First, an example of Pareto optimum that cannot be sustained as a compet-
itive equilibrium is that
√ √
uA (x1 , x2 ) = x1 + x2 , ωA = (0, 1)
uB (x1 , x2 ) = x1 , ωB = (1, 0)
Here, the initial allocation is Pareto optimal, but there is no price vector that
sustains it as a Walrasian equilibrium. The problem is that B’s preferences are
not strongly monotone.
Another example of Pareto optimum that cannot be sustained as a Walrasian
equilibrium is that consumers’ preferences are nonconvex.
An example of Walrasian equilibrium that is not Pareto optimal is one with
locally satiated preferences.
An example of no competitive equilibrium, given initial endowments, is the
first example.
With perfect complements, we can have an infinite number of equilibria.
The same allocation can be supported by an infinite number of price ratios.
We can also encounter multiple equilibrium allocations with not strictly
convex preferences, for instance, perfect substitutes.
The Walrasian equilibrium and Pareto optimality concepts say nothing about
fairness. An allocation can be Pareto optimal while being very unfair. For in-
stance, giving all the resources of the economy to one of the consumers is Pareto
optimal, but nobody would say it is fair (except of, possibly, that particular con-
sumer).
4 Pareto Frontier and Walras’ Law
The set of Pareto optimal allocations in the economy is characterized by the
Pareto frontier. This utility possibilities frontier is characterized by
U P = {(u1 , ..., uI ) ∈ U : @(u01 , ..., u0I ) ∈ U s.t. u0i < ui ∀i and u0i ui for some i},
where these levels of utility are attained by feasible allocations x, y. These al-
locations can be characterized by maximizing the welfare of a consumer, say 1,
subject to feasibility and to the fact that the remaining consumers get at least
some level of utility:
maxu1 (x11 , ..., xL1 )
x,y
s.t. u2 (x12 , ..., xL2 ) ≥ u2 i = 2, ..., I
X X
xli ≤ ω l + yli j = 1, ..., J
i j
Fj (y1j , ..., yLj ) ≤ 0 j = 1, ..., J
If the Pareto frontier is convex, we can characterize all Pareto optimal allo-
cations by means of a linear social welfare function
2
X
W (u1 , ..., uI ) = λi ui ,
i
and every Pareto efficient allocation corresponds to some weight of the different
consumers’ utility functions. In other words, if an allocation maximizes some
social welfare function, then it is Pareto efficient.
Define the excess demand function as
I
X
z(p) = [xi (p, pω i ) − ωi ]
i=1
Walras’ Law states that for any price vector p, we have pz(p) ≡ 0, i.e. the
value of the excess demand is identically zero. This is because each consumer’s
budget constraint must be satisfied with equality. A corollary of Walras’ Law is
that if demand equals supply in k − 1 markets and pk > 0, then demand equals
supply in the kth market.
Furthermore, if all goods are desirable and p∗ is a Walrasian equilibrium,
then z(p∗ ) = 0.
The proof of existence of a Walrasian equilibrium relies on Brouwer’s fixed
point theorem. If z : S k−1 → Rk is a continuous function that satisfies Walras’
Law, then there exists some p∗ in S k−1 such that z(p∗ ) ≤ 0.
Regarding uniqueness, if all goods are gross substitutes at all prices, then if
p∗ is an equilibrium price vector, it is the unique equilibrium price vector.
Two goods, i and j, are gross substitutes at a price vector p if
∂zj (p)
> 0, for i 6= j.
∂pi
