ECON600 2022
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
1. In this problem, we consider a decision maker who must solve a two period investment problem. In each period, exactly one of two risky investments will be available. The decision maker has w > 0 dollars of investable wealth and a thrice continuously di§erentiable von Neumann -Morgenstern utility function u : R ! R satisfying u\ (x) > 0 and u\\ (x) < 0 for all x 2 R. In period i, the asset return is the realization of a nonnegative random variable Yi . Only asset i is available in period i and the investor must commit to his investment decisions before learning the realizations of the asset returns. Let x denote investment in period 1 and w - x denote investment in period 2 where x 0 and 0 < x < w .
Consequently, his optimization problem may be formulated as
maximize E[u(xY1 )+ u((w - x)Y2 )] subject to 0 < x < w:
Let x denote the optimal solution and assume that 0 < x < w: Suppose that Y1 %C Y2 (i.e., Y1 dominates Y2 in the concave order) and that the the investorís
Arrow-Pratt measure of prudence is nonpositive for all wealth levels. That is, u\\\ (x) < 0 for all x. Show that x w - x:
Solution: This problem is essentially identical question 1 on the 2019 com- prehensive exam. Let
f(x) = E[u(xY1 )+ u((w - x)Y2 )]
and note that f\ (x) = 0: Next, note that f\\ (x) < 0 for all x. Therefore, to show that x ; it su¢ces to show that f\ ( ) 0: To see this note that
w w w w w
2 2 2 2 2
DeÖning '(y) = u\ ( y)y; note that ' concave will imply that
w w w
2 2 2
since Y1 %c Y2 :
Computing the derivatives, we obtain
'\ (y) = u\\ ( y) ( y) + u\ ( y)
'\\ (y) = u\\\ ( y) y) + u\\ ( y)w
Consequently, f\ ( ) 0 if
u\\\ ( y) y) + u\\ ( y)w < 0:
2. Consider a pure exchange economy with L goods and n consumers. Agent i has utility function ui : R ! R and endowment !i 2 R . Let N = f1;::;ng and deÖne the set F(N) of feasible allocations as
F(N) = f(xi )i2N :X xi <X !i and xi 2 R for each i 2 N:
i2N i2N
Various criteria have been proposed for identifying strongly Pareto optimal al- locations. Let V (N) denote the set of attainable payo§s ssociated with this economy. Formally, let
V (N) = fv 2 Rn j(v1 ;::;vn ) = (u1 (x1 );:::;un (xn ) for some (xi )i2N 2 F(N)g:
Suppose that g : Rn ! R is strongly monotonic: if x y and x y; then g(x) > g(y). Suppose that (v 1 ;::; vn ) is a solution to the optimization problem
maximize g(v1 ;:::;vn )
st (v1 ;:::;vn ) 2 V (N):
If (x1 ;::; xn ) 2 F(N) and (v 1 ;::; vn ) = (u1 (x1 );:::;un (xn ); show that the alloca- tion (x1 ;::; xn ) is strongly Pareto optimal.
Solution: Suppose that (v 1 ;::; vn ) is a solution to the optimization prob- lem. Suppose that (x1 ;::; xn ) 2 F(N) and (v 1 ;::; vn ) = (u1 (x1 );:::;un (xn ): If (x1 ;::; xn ) is not strongly Pareto optimal, then there exists a feasible allocation (y1 ;::;yn ) 2 F(N) such that ui (yi ) ui (xi ) for each i and ui* (yi* ) > ui* (xi* ) for some i : Note that (u1 (y1 );:::;un (yn )) 2 V (N): Since g is strongly monotonic, it follows that
g(u1 (y1 );:::;un (yn )) > g(u1 (x1 );:::;un (xn )) = g(v 1 ;::; vn )
but this is impossible since (v 1 ;::; vn ) is a solution to the optimization problem.
3. In this problem, we consider a decision maker who must solve a two period consumption-investment problem. The decision maker has w > 0 dollars to allocate for investment and immediate consumption. In addition, the decision maker has a twice continuously di§erentiable von Neumann -Morgenstern utility function u : R ! R satisfying u\ (x) > 0 and u\\ (x) < 0 for all x 2 R. Let w x denote immediate consumption in period 1 and let x denote investment where 0 < x < w . In addition to the choice of x; the consumer will have additional income in period 1 determined as the realization of a nonnegative random variable. In period 1, this random variable is Z . In period 2, the random return per dollar invested can be either Y1 or Y2 . The problem is made complicated by the fact that the individual must commit to his consumption decision w x and investment decision x before learning the realizations of the random variables in each period. If the random second period return is Yi ; then his optimization problem may be formulated as
maximize E[u(w x + Z)+ u(xYi )]
st 0 < x < w:
Let xi denote the optimal solution when the second period return is Yi and assume that 0 < xi < w: Suppose that R(z) = for all z . Furthermore, suppose that Y2 is a mean preserving spead of Y1 : In particular, suppose that
the random vector (Y1 ;Y2 ) has a joint density f with marginals fY1 and fY2
such that E[Y2 jY1 = y] =R01 y2 f(y2 jy1 )dy2 = y1 for all y1 0: If R(z) < 1 for
all z and if R is a di§erentiable, increasing function of z, show that x1 x2 :
Solution: This question is identical to problem 3 on the May 2019 exam.
In that question, it was assumed that Y1 %C Y2 while this question makes the
equivalent assumption that Y2 is a mean preserving spread of Y1 : Let
fi (x) = E[u(w - x + Z)+ u(xYi )] = E[u(w - x + Z)] + E[u(xYi )]:
Then f(x) < 0 for all x and f(xi ) = 0 for each i. As usual, we wish to show
that f(x1 ) < 0: To see this, deÖne
'(y) = -E[u\ (w - x1 + Z)] + u\ (x1 y)y
Then
'\ (y) = u\\ (x1 y)(x1 y)+ u\ (x1 y)
In addition, 1 - R(z) 0 and R\ (z) 0 imply that
'\\ (y) = u\\ (x1 y)x1 [1 - R(x1 y)] - u\ (x1 y)R\ (x1 y)x1
< 0:
Since Y2 is a mps of Y1 ; we conclude that
0 = f(x1 )
= E['(Y1 )]
1
Z 1 Z1 )
Z0 1 Z0 1 '(y2 )f(y2 jy1 )dy2 ) fY1 (y1 )dy1
= Z0 1 '(y2 ) Z0 1 f(y2 jy1 )fY1 (y1 )dy1 ) dy2
= Z0 1 '(y2 ) Z0 1 f(y1 ;y2 )dy1 ) dy2
= Z0 1 '(y2 )fY2 (y2 )dy2
= E['(Y2 )]
= f(x1 ):
2023-06-30