Solution To Problem Set Three - Choice
Solution To Problem Set Three - Choice
Solution To Problem Set Three - Choice
(a) The decision maker has in mind a ranking of all alternatives and
chooses the alternative that is the worst according to this ranking.
Yes. Let % be the preference relation reflecting the ranking of the
decision maker (DM). This choice procedure is rationalized by the
preference relation %∗ , where b %∗ a whenever a % b.
(b) The decision maker chooses an alternative with the intention that
another person will suffer the most.
Yes. The argument is analogous to (a) once we define % to be the
preference relation reflecting the ranking of the other person.
(c) The decision maker asks his two children to rank the alternatives and
then chooses the alternative that is the best “on average.”
The answer depends on the interpretation of “best on average:”
Yes: If the DM calculates the “best on average” over a (finite) set X,
then his procedure is rationalizable. Assume that both children
have well defined preferences over the alternatives, and that both
assign a number vi (x) ∈ {1, ..., n} to each element x ∈ X with
the interpretation that x %i y iff vi (x) ≥ vi (y). The DM will
thus have a complete and transitive ranking over X such that
x % y iff v1 (x)+v
2
2 (x)
≥ v1 (y)+v
2
2 (y)
.
No: If the DM calculates the “best on average” for each choice prob-
lem A ⊆ X, then the procedure need not be rationalizable: Let
X = {x, y, z} and the children’s rankings be
x %1 y %1 z and y %2 z %2 x.
1
behavior can be rationalized by the preferences represented by the
utility function u(a) = −d(a, x).
(e) The decision maker looks for the alternative that appears most often
in the choice set.
No. Note that C({a, a, b}) = a 6= b = C({a, b, b}). Since {a, a, b} and
{a, b, b} are the same set, C is not rationalizable. See Question 6 for
more on choice from lists (as opposed to sets).
(f) The decision maker always selects the first alternative that comes to
his attention.
No. Consider a DM that chooses the first alternative in a set. Note
that C({a, b}) = a 6= b = C({b, a}). See (e) for discussion.
(g) The decision maker searches for someone he knows who will choose
an action that is feasible for him.
No. See (e) and (f).
(h) The decision maker orders all alternatives from left to right and se-
lects the median.
No. Note that C({a, b, c}) = b 6= c = C({b, c, a}). See (e), (f) & (g).
2. Let us say that you have to make a choice from a set A. Does it matter
whether (a) you make a choice from the entire set or (b) you first partition
A into the subsets A1 and A2 , then make a selection from each of the sets
and finally make a choice from the elements you selected from among A1
and A2 ?
(b) Show that the rational decision maker satisfies this property.
Let % be the preference relation of our rational DM and A1 , A2 be
a partition of A.
2
Next, choose any x ∈ C% C% (A1 ) ∪ C% (A2 ) , and note x % y for
every y ∈ C% (A1 ) ∪ C% (A2 ). Without loss of generality, assume
that x ∈ C% (A1 ) which implies x % a for every a ∈ A1 . Moreover,
y ∈ C% (A2 ) implies that x % y % b for every b ∈ A2 . Combining
these results, we have x % c for every c ∈ A1 ∪ A2 = A, and thus
x ∈ C% (A) ⇒ C% C% (A1 ) ∪ C% (A2 ) ⊆ C% (A).
(c) Find examples of choice procedures that do not satisfy this property.
i. Order the alternatives in A from left to right and choose the
second alternative if |A| ≥ 2 and the first alternative otherwise.
Note that C({x, y, z}) = y 6= z = C(C({x, y}) ∪ C({z})).
ii. The choice procedure outlined in 3(b). Let X = {a, b, c, d, e}
and V (a) > V (b) > V (c) > V (d) > V (e). Note that C(X) =
{a, b, c} =
6 {a, b, d} = C(C({a, b, c} ∪ C({d, e})).
(d) Show that if a (single-valued) choice function satisfies path indepen-
dence, then it is consistent with rationality.
Let C be a single-valued choice function that satisfies path indepen-
dence. Choose any sets A ⊂ B ⊆ X such that C(B) ∈ A. Path
independence implies that C(B) = C C(A) ∪ C(B \ A) . Since
C(B) ∈/ B \ A, then this implies that C(B) = C(A).
C, therefore, satisfies Condition * for all sets of subsets of X (includ-
ing those of size three or fewer), which implies that that C can be
rationalized.
(e) Assume that C is a (multi-valued) choice function satisfying path
independence. Can C necessarily be rationalized?
No, not necessarily. Consider the following counterexample:
• C({x, y, z}) = {x, y}.
• C({x, y}) = {x, y}; C({x, z}) = {x, z}; C({y, z}) = {y}.
• C({x}) = {x}; C({y}) = {y}; C({z}) = {z}.
Note that C satisfies path independence (verify this!) but fails the
WA since z ∈ C({x, z}) and x ∈ C({x, y, z}), but z ∈
/ C({x, y, z}).
3
(b) D(A) = {x ∈ A | the number of y ∈ A for which V (x) ≥ V (y) is at
least |A|
2 }.
No, D does not satisfy the WA. Consider the following counterex-
ample: Let X = {a, b, c, d, e} be such that V (a) > V (b) > V (c) >
V (d) > V (e), and let A = {a, b, c}. Note that a, c ∈ A, c ∈ C(X)
and a ∈ C(A), but c ∈/ C(A).
4
5. Consider the following two choice procedures. Explain each procedure and
convince a skeptic why each “makes sense.” For each, determine whether
they are consistent with the “rational man” paradigm.
(a) The primatives of the procedure are two numerical (one-to-one) func-
tions u and v defined on X and a number v ∗ . For any given choice
problem A, let a∗ ∈ A be the maximizer of u over A, and let b∗ ∈ A
be the maximizer of v over the set A. The decision maker chooses a∗
if v(a∗ ) ≥ v ∗ and chooses b∗ if v(a∗ ) < v ∗ .
The following environment exemplifies the procedure: Consider a
DM that is shopping for a car. Let X represent the set of cars on
the market and A represent the cars in X that the DM can afford.
Assume the DM ranks the cars on two criterion: by fuel efficiency,
represented by u, and safety, represented by v. The DM will thus
choose the most fuel efficient car in A if it has a safety rating of at
least v ∗ , and otherwise will choose the safest car in A.
While this procedure seems reasonable, it nevertheless fails the “ra-
tional man” paradigm: Let X = {x, y, z}, v ∗ = 2 and the “values”
of the alternatives be:
Element u(.) v(.)
x 3 1
y 2 2
z 1 3
Note that the DM will choose z from X but will choose y from {y, z},
thus violating the WA.
(b) The primatives of the procedure are two numerical (one-to-one) func-
tions u and v defined on X and a number u∗ . For any given choice
problem A, let a∗ ∈ A be the maximizer of u over A, and let b∗ ∈ A
be the maximizer of v over the set A. The decision maker chooses a∗
if u(a∗ ) ≥ u∗ and chooses b∗ if u(a∗ ) < u∗ .
Consider the example presented in (a): The DM will now choose the
most fuel efficient car in A if it attains a fuel efficient rating of at
least u∗ , and will choose the safest car in A otherwise.
This procedure is rationalizable: Define the set
G = {z ∈ X | u(z) ≥ u∗ }.
Let A, B ⊆ X and x, y ∈ A∩B be such that x ∈ C(A) and y ∈ C(B).
First, assume that x ∈ G, which implies that u(x) ≥ u(y) since y ∈ A.
Moreover, since x ∈ B then u(y) ≥ u(x) ≥ u∗ , which implies that
u(x) = u(y), and consequently x ∈ C(B).
Next, assume x ∈ / G, which implies that u(a) ∈ / G for every a ∈ A
since x ∈ C(A). Consequently, v(x) ≥ v(y). Moreover, since y ∈ /G
and y ∈ C(B), then u(b) ∈/ G for every b ∈ B. Therefore, v(y) ≥ v(x).
Consequently, v(x) = v(y) which implies x ∈ C(B).
5
6. The standard economic model assumes that choice is made from a set. Let
us construct a model where choice is assumed to be made from a list. Let X
be a finite “grand set.” A list is a nonempty, finite vector of elements in X.
In this problem, consider a choice function C to be a function that assigns
to each vector L =< a1 , . . . , ak > a single element from {a1 , . . . , ak }.
Thus, for example, the list < a, b > is distinct from < a, a, b > and <
b, a >. For all L1 , . . . , Lm define < L1 , . . . , Lm > to be the list that is the
concatenation of the m lists. Note P that if the length of Li is ki , then the
length of the concatenation is i=1,...,m ki . We say that L‘ extends the
list L if there is a list M such that L‘ =< L, M > .
We say that a choice function C satisfies property I if for all L1 , . . . , Lm ,
we have C(< L1 , . . . , Lm >) = C(< C(L1 ), . . . , C(Lm ) >).
(a) Interpret property I. Give two examples of choice functions that sat-
isfy I and two examples of choice functions which do not.
Property I is analogous to the path independence property from ques-
tion 2. C satisfies I if the choice made from a grand list L is the same
as first partitioning L into smaller lists L1 , ..., Lm which preserve the
original order of L, choosing an alternative from each Li and then
making a subsequent choice from the list < C(L1 ), ..., C(Lm ) >.
Two Choice Functions Satisfying I :
• The DM chooses the first element from the list.
• Let u : X → < represent the utility the DM receives from
choosing an alternative in X, and let u∗ ∈ < represent some
utility threshold. The procedure whereby the DM chooses
the first alternative in the list that yields utility of at least
u∗ and the last alternative if none of the elements yield u∗
also satisfies I.
Two Choice Functions that Fail I :
• The DM chooses the second alternative from the list if |L| ≥ 2
and the first alternative otherwise.
• Again, let u : X → < represent the utility the DM receives
from choosing an alternative, and let u∗ ∈ < represent some
utility threshold. The procedure whereby the DM chooses
the first alternative in L if it yields utility of at least u∗ and
the last alternative otherwise fails I.
(b) Define formally the following two properties of a choice function:
• Order Invariance: A change in the order of the elements of
the list does not alter the choice.
Let L =< a1 , . . . , aK >. A permutation of L is a list L =<
aπ(1) , . . . , aπ(K) >, where π : {1, . . . , K} → {1, . . . , K} is a one-
to-one, onto function. The choice function C satisfies Order
Invariance if C(L) = C(L) for every permutation L of L.
6
• Duplication Invariance: Deleting an element that appears in
the list elsewhere does not change the choice.
C satisfies Duplication Invariance if C(L) = C(L0 ) for any L,
L0 such that L =< M, a, M 0 > and L0 =< M, M 0 >, where a is
an alternative in either M or M 0 .
(c) Characterize the choice functions that satisfy Order Invariance, Du-
plication Invariance and Condition I.
Define: For every list L =< a1 , ..., am >, define L = {a1 , ..., ak }
to be the set of the alternatives in L. For every (finite) set S =
{a1 , ..., ak } ⊆ X, define S 0 =< a1 , ..., ak > to be the list of all ele-
ments in S, each appearing once in some arbitrary order.
Choose any two lists K, L such that K = L, and denote that the
cardinality of this set by n. To show that C is well defined, we must
prove that C(K) = C(L). Define the permutations πK , πL such that
KπK =< K ∗ , MK > and LπL =< L∗ , ML >, where K ∗ and L∗ are
lists of length n that contain precisely the n different alternatives.
Since K = L, then K ∗ and L∗ contain the same alternatives.
By OI note that C(K) = C(KπK ) and C(L) = C(KπL ), and by DI
note that C(KπK ) = C(K ∗ ) and C(LπL ) = C(L∗ ). Moreover, since
there exists a permutation π 0 such that Kπ∗0 = L∗ , OI implies that
C(K ∗ ) = C(L∗ ), and consequently C(K) = C(L). Moreover, since
C is single-valued, this implies that C is single-valued as well.
7
(d) Assume now that in the back of the decision maker’s mind is a value
function u defined on the set X such that u(x) 6= u(y) ∀x 6= y.
For any choice function C define vC (L) = u(C(L)). We say that
C accommodates a longer list if whenever L0 extends L, vC (L0 ) ≥
vC (L) and there exists a list L0 which extends a list L for which
vC (L0 ) > vC (L). Give two interesting examples of choice functions
that accommodate a longer list.
• The DM chooses the u-maximal element in L.
• The DM chooses the first element in L if it is the u-maximal
alternative, and chooses the alternative that yields the second
highest value of u otherwise.
(e) Give two interesting examples of choice functions which satisfy prop-
erty I but which do not accommodate a longer list.
• The DM chooses the u-minimal element in L.
• The DM always chooses the first element in the list.