Soluções site Jehle Reny

Capítulo 1

1.5

a) Let x be in ~(x0). Then x>=x0 and x0>=x so x is in >=(x0) intersection <=(x0).

Let x be in >=(x0) intersection <=(x0). Then x>=x0 and x0>=x so x~xo and x is in ~(x0).
(c) Assume there exists x in ~(x0)intersection >(x0). Then there exists an x such that x~x0
and x>x0, a contradiction.
(e) Clearly X = >=(x0) union x<=(x0). Take the complement.

1.13

(a) You are justified to be concerned. As stated in the book, we have a problem with
reflexivity (a bundle is at least as preferred as itself) and the indifference set being empty.
This was pointed out in an earlier posting by Bettina Klaus. The usual definition of a
lexicographic ordering is in strict preference if strict inequalities are used. With this definition
the following two quotes may be helpful. 'The bundles are ordered according to the principle
of words in a dictionary or lexicography...In this case, each bundle has no points (other than
itself) to which it is indifferent; indifference surfaces cannot be drawn and no utility function
exists'. (Deaton and Muellbauer, Economics and consumer behavior (1980)). Or look up the
entry by C Blackorby on lexicographic orderings in The New Palgrave: 'There are no trade-
offs between [bundles] a and b and each indifference set is a single point'.
So the answer is a single point using the usual definition of a lexicographic ordering.
(b) For a proof see Debreu (1954), footnote 1 as referenced on page 531 of the book. The
paper is also in Debreu's book 'Mathematical Economics'. Although the footnote is short, it is
not easy! The proof in Mas-Colell, Whinston and Green 'Microeconomic Theory' (page 46) is
easier to follow.

1.15

The budget set, B, is defined on page 20. In the book a compact set in Rn is defined as a set
that is closed and bounded (p.426).

'Closed' means all the boundary points are in the set (p.422). I hope this is obvious for B,
but if not try drawing a diagram. A more technical answer is that the sets X={x:x in Rn+}
and Y={x:px<=y} are both closed so B=(X intersection Y) is closed by Theorem A1.4.4
(p.423).

'Bounded' means the set is enclosed in a closed ball (p.423). Note that p>>o means that q =
min(p1,p2,...,pn)>0. Now take the closed ball with centre 0 and radius y/q. The budget set
is enclosed in this ball. Draw a diagram again!

So we have shown compactness.

For convexity, take w and x in B and show that aw + (1-a)x is also in B where 0<=a<=1
(p.412).

1.20

Forming the Lagrangian:

L = Ax1x21- + (y-p1x1-p2x2)

F.O.C:

L/x1 = Ax1-1x21- - p1 = 0 (1)

L/x2 = A(1-)x1x2- - p2 = 0 (2)

L/ = y-p1x1-p2x2 = 0 (3)

Dividing (1) by (2) we obtain,

( / 1-)  ( x2 / x1 ) = (p1/ p2)

 x2 = (1- )/  (p1/ p2)  x1 (4)

Substituting (4) into (3) we obtain:

y = p1x1+p2  (1- )/  (p1/ p2)  x1

= (p1x1)/

 x1 = (/p1)y

Similarly:
  x2 = y(1-)/p2

1.25

Here's my hint: First, let's restrict attention to the strictly positive orthant.
If preferences are monotonic there, then one thing we know is that utility can not go down
as consumption of either x_1 or x_2 increases, holding the other constant. Differentiate
u(x_1,x_2) with respect to x_1, and do the same with respect
to x_2, and see what this tells you must be true of the value of alpha.

That's really only part of the picture, though. We know that convex monotonic preferences
are represented by increasing and quasiconcave utility functions. Hence, we need to check if
quasiconcavity requires anything additional of alpha. One way to do that is to apply the test
for quasiconcavity given in Exercise A2.18 of the text. (Unfortunately, Joon, in the edition of
the text you have, there is an error in the description of that test that is given in exercise
2.18. The error is corrected in the new edition, so I have attached a pdf file containing the
corrected version to this reply for you to consult. It is called "Arrow_Enthoven") When you
apply that test, however, you'll see nothing more is needed than what you will have deduced
above.

1.25

A consumer with convex and monotonic preferences. Consumes X 1 & X2 0

His Utility function is given by U (X1 , X2) = X1.X2(1/2) -

a. Restrictions on the parameter ?

The overall utility is supposed to increase as the consumptions of X 1 and X2

U/X X1-1. X2(1/2 -  0

 

U/X2 = (1/2 -.X1.X2 –1/2 -  0

 (1/2 -      (2)

From (1) and (2)     which is the restriction on .

__________________________________

b. Marshallian demand functions?

To calculate the Marshallian demand functions ( X 1* , X2*) we use the Lagrange

Multiplier:

L = X1.X2(1/2) - +  .X1 .X2

First Order Condition (F.O.C) 

L/X1 = .X1(. X2(1/2) -   (3)

L/X2 = (1/2 - ).X1. X2-1/2 - 2  (4)

Y – P1.X1 – P2.X2 = 0 (5)

Dividing (3) by (4) we obtain,

.X2 / ( - .X1 = P1 / P2

 X2 = (1/2 -   . X1 (6)

Substituting (6) into (5), we obtain,

X1* = 2   1

X2* = (1 - 2   

1.27

The hint in the text suggests you sketch the indifference map. When you do, you will find
that the indifference curves are "V-shaped", and convex away from the origin. The Vee's all
kink along the line x_2=x_1: above that line all indifference curve segments have a slope of
-1/a, while below the kink their slope is -a. (Note that because 0 < a < 1, the former is
absolutely steeper than the latter, forming the Vee.)

Now suppose any level of income, and ask yourself the following questions: what happens
when p1/p2 < a? How about when
a< p1/p2 < 1/a ? How about when 1/a < p1/p2? Answer all these, and resolve what happens
when equality holds in the ranges I've indicated, and you will have described Marshallian
demand at all relative prices.
1.28 An infinitely lived agent owns 1 unit of a commodity that she consumes over
her lifetime. The commodity is perfectly storable and she will receive no more
than she has now. Consumption of the commodity in period t is denoted xt ,
and her lifetime utility function is given by

U(x0,x1,x2,……)=+t ln (xt), where 01.

t=0

Calculate her optimal level of consumption in each period.

1. Forming the Lagrangian:

+t ln (xt) + (1-+ln xt )
t=0 t=0

L = t -  =0

xt xt

L = t+1 -  =0

xt+1 xt+1

2. Set the two ’s equal to each other.

xt+1=  xt where 01

x1=  x0

x2=  x1= 2 x0

xt= tx0
 3. Substitute into budget constraint
1= +xt = x0 (1+  +….+ t+….)
t=0

= x0

1-

x0 = (1-)

x1 =  (1-)

x2 = 2 (1-)

xt = t (1-)

Thus it converges.

THE END

1.29

Slope:- Holding v and y constant, differentiate totally to get dp2/dp1= -Dv/Dp1/Dv/Dp2<0.

(D denotes partial differentiation.) Theorem 1.6.4. Direction of increasing utility:- Decreasing
p1 and p2. Theorem 1.6.4 Curvature:- Pick two points on the price-indifference curve and
use quasiconvexity to show the curve is bowed towards the origin. Theorem 1.6.5. 1.46 (a)
e(p,1)=min px subject to u(x)>=1 so, using homogeneity of u(x),
e(p,1)=1/u* min pu*x subject to u*u(x)=u(u*x)>=u*.
Substituting z=u*x gives u*e(p,1)=min pz subject to u(z)>=u* and this equals e(p,u*).
(b)e(p,u)=ue(p,1) and so de(p,u)/du=e(p,1).

1.33

I see what you are getting at but unbounded utility is not needed for the result as the
following counterexample shows.

Consider the utility function u(x1,x2) = 1 - 1/(x1x2) which is continuous, strictly increasing
in x1 and x2, bounded above by 1 and has expenditure function e(p1,p2,u) =
2*sqrt[p1p2/(1-u)]. This is unbounded above as u tends to 1.

To answer the question, consider what would happen if expenditure were bounded above and
the consumer were given an income above this bound. Some portion of income would not be
spent but this is impossible as we know the consumer always wants more.

To put it formally, suppose for p>>0 that there exists a y* such that e(p,u)<= y* for all u.
Now choose y** > y* and denote the maximum utility achievable with income y** by u**.
Since the utility function is continuous and strictly increasing, by Theorem 1.8 we have
e(p,u**) = e(p,v(p,y**)) = y** <= y* < y**, a contradiction.

1.46
a) e(p,1)=minP*X u(x)>=1
u*e(p,1)=u*minP*X u*u(x)>=u*1=u
u(x) is a linear homogeneous utility function
so u*u(x)=u(ux)>=u
u*e(p,1)=u*minP*X=minP*uX=e(P,u),u(ux)>=u
so e(P,u)=e(P,1)*u
(b) according to (a),e(P,u)=e(P,1)*u,
and e(P,u)=y,so y=e(P,1)*u,
then, du/dy=1/e(P,u)
so,the marginal utilty of income depends on P,but is independent of y.

1.47

We have e(p,u) = k(u)g(p) which is in the "Hicks" variables (p,u) whereas the income
elasticity is defined on the "Marshall" variables (p,y). I shall use the Theorems connecting
Hicks and Marshall together with the Chain Rule.
From Theorem 1.7.7 p.43,
de(p,u)/dp = xh(p,u)
so xh = kdg/dp
and dxh(p,u)/du = k'dg/dp.
From Theorem 1.9.1 p.43,
x(p,y) = xh(p,v(p,y))
and so x = kdg/dp (*).
Also dx/dy = dxh/du.dv/dy.
From Theorem 1.8.1 p. 40, e(p,v(p,y)) = y so de/du.dv/dy = 1 and since
de/du = k'g,
dv/dy = 1/(k'g).
Thus dx/dy = (1/g)dg/dp (*).
Also y = kg (*).
Income elasticity equals y/x.dx/dy and the result follows on substituting the expressions
marked (*).

1,49

1.49(a), you make an interesting point: if we take the phrase "consumes a single good" to
mean that the consumer consumes only the good $x$, then you rightly observe that the
income elasticity of demand for $x$ must be unity. However, that's not what we intended
that phrase to mean.

Instead, think of $x$ as one of many goods in the consumer's budget, but let it be the
"single good" on which we want to focus our attention. Writing indirect utility as a function of
that good's price and income, alone, is a (perhaps confusing?) way to indicate that we intend
to focus on demand for that good alone, assuming that the prices of all other goods remain
fixed.

Thus, for example, if the single good of interest is good 1, the notation v(p,y) should be
viewed as a short-hand for
v(p_1,p_2^0, p_3^0...p_n^0,y) where the "^0" indicates fixed values for the other goods'
prices.
b) Use Exercise 1.32 to transform the indirect utility function to
v*(p,y) = G inverse (v(p,y)).
Then v*(p',y bar) - v*(p,y bar) = A(p') - A(p).
Split the integral using p < p' < p zero to get the result.
The utility loss is the negative of an area under a demand curve and this is money.
But I have a problem with 1.49(a). There is only one good so we can immediately write down
the demand function as x(p,y) = y/p and this has an income elasticity of one. But this
messes up the interpretation of nu.
Puzzled? So am I!

1.52

What you want to show is whether or not the Marshallian demands are the same given the
conditions on the expenditure functions. Marshallian demands, x(p,y), are observable;
Hicksian demands, h(p,u), are not.
(a) eA(p,2u)=eB(p,u)=y so hA(p,2u)=hB(p,u) using Shephard's lemma. Hence
xA(p,eA(p,2u))=xB(p,eB(p,u))as x(p,e(p,u))=h(p,u) by Theorem 1.9. Thus xA(p,y)=xB(p,y).
(b) In a similar fashion to part (a), 2xA(p,y)=xB(p,2y). Now use the result of Exercise 1.47
to get xA(p,2y)=2xA(p,y) and then deduce that xA(p,y)=xB(p,y).

1.54

This concerns preferences which are additively separable. The cross partial derivatives of the
utility function are zero. I think this structure of preferences is known as a Gossen map and
it has some rather odd characteristics as we are about to find out.
(a) Suppose good 1 displays increasing marginal utility so f1''>0. Using the result on page
511, D3=f1'f1'fj''+fj'fj'f1''<0 so fj''<0 all j.
(b) From the first order conditions, fi'(xi(p,y))=L(p,y).pi where L(p,y) is the Lagrangian
multiplier. Differentiate with respect to y using the Chain Rule to get fi''(xi).dxi/dy=dL/dy.pi
and so (fi''.dxi/dy)/pi=(fj''.dxj/dy)/pj all i and j(**). Now p>>0 so if f1''>0 and fj''<0 all j not
equal to 1, then I. EITHER if dx1/dy>0 (good 1 normal) then dxj/dy<0 (good j inferior) all j
not equal to 1 II. OR if dx1/dy<0 (good 1 inferior) then dxj/dy>0 (good j normal) all j not
equal to 1.
I cannot think why the second case should be ruled out. It would seem to be the 'usual' case
as it says that inferior goods are rare.
(c) If fj''<0 all j then either dxj/dy<0 all j or dxj/dy>0 all j using (**). But p.x=y so
p.dx/dy=1 and since p>>0 we cannot have dxj/dy<0 for all j. Hence dxj/dy>0 all j and all
goods are normal.
An excellent answer to 1.57 has already been posted. The first step is to write u=v(p,y) and
y=e(p,u), re-arrange to get v(p,y)=a*(p)+b*(p)y where a*(p)=-a(p)/p and b*(p)=1/b(p).
(Introduce new notation to save mess! I hope my old tutor Terence Gorman would have
approved.) Then use Roy's identity to get xi(p,y), then calculate dxi/dy and get that the
income elasticity of demand equals bi*(p)y/(ai*(p)+bi*(p)y) where the subscript denotes
differentation with respect to pi.

1.55

This replaces a message about Q1.55(b) that I posted earlier. My first answer was a bit too
cryptic and incomplete. d denotes partial differentiation.
We have v(p1,p2,y)=w(p1,p2)+z(p1,p2)/y and p1>0, p2>0 and y>0. (Since we are dealing
with zero demand if y=0 and p1>0 and p2>0, I feel justified in altering the question a little!
It also removes the temptation to divide by zero.)
We need to find conditions on w and z to satisfy conditions 1 to 5 of Theorem 1.6 on page
28.
Condition 3 requires that v is strictly increasing in y, so z<0.
It is a bit difficult to make much further progress without assuming that both w and z are
continuous and differentiable in (p1,p2) so we shall do so. Condition 1 is met by this
assumption.
Condition 4 requires that v is decreasing in p1 and p2, so dw/dp1+dz/dp1.y<=0 (*). Letting
y tend to zero gives dw/dp1<=0. Similarly dw/dp2<=0. Now divide(*)by y>0 and let y tend
to infinity to get dz/p1<=0 and similarly dz/dp2<=0.
Condition 2 requires that v is homogeneous of degree zero in (p1,p2,y). So
v(p1,p2,y)=v(tp1,tp2,ty) for all t>0. Write this out on full in w and z, differentiate with
respect to t using the Chain Rule and the Quotient Rule, set t=1 and you should get the
following expression (**):
0=dw/dp1.p1+dw/dp2.p2+{dz/dp1.p1+dz/dp2.p2-z}/y. Now let y tend to infinity to get that
dw/dp1.p1+dw/dp2.p2=0. But condition 4 implied that dw/dpi<=0 for i=1,2 so dw/dpi=0 for
i=1,2 as pi>0. Thus w(p1,p2)=k, a constant. (Check this!)
Now go back to (**), multiply through by y>0 and let y tend to zero to get
dz.dp1.p1+dz/dp2.p2=z. Euler's Theorem (p.471) implies that z is homogeneous of degree 1
in (p1,p2).
Condition 5 implies that z is quasiconvex in (p1,p2). You can check there is no further
restriction implied by v being quasiconvex in (p1,p2,y). (Remember that z<0).
I am happier with this answer than my earlier attempt.
Finally, you may think that Condition 1 has simply been assumed away and you are right!
Although it is true that if f and g are continuous functions then f+g is also continuous, it is
easy to think up functions f and g which are not continuous whereas f+g is continuous. For
example, paste together two step functions. More exotically, consider f(x)=1 for x rational
and f(x)=0 for irrational x together with g(x)=0 for x rational and g(x)=1 for irrational x.
Neither f nor g is continuous anywhere but f+g=1. These are Dirichlet functions, if you want
to look up the details.

1.57

1. Give e(p,u), we can find v(p,y).

2. Use Roy's Identity to find x(p,y).
3. Differentiate x(p,y) w.r.t y and multiply the result by y/x to get income elasticity.
4. The result is a bit messy but don't worry!
5. Divide numerator and denominator by y.
6. Take limit as y approaches o and infinity, respectively.
7. Done!!!
Hope this helps.

1.62

The problem is straight forward.

We know that the substitution matrix, S, is symmetric. Therefore, b = 2. Furthermore, S is

negative semidefinite. So we know that the sign of every odd order principal minor is non-
positive and every even order principal minor is non-negative. This then implies that a =< -
8, but we do not know its exact value yet.

From here it gets a little trickier. First, we know that the Hicksian demand function is
homogeneous with degree 0 in prices (this follows from expenditure minimization problem).
By Euler's Theorem, we know then that S * P = 0, where P is a price vector. Because we
know that the price of good-one is 8, we have two equations and two unknowns, namely p
and a. Hence, we have p = 32 and a = -8.
So we have: a = -8, b = 2, and p = 32.

1.61

(a) use the fact that e(p,u) is homogeneous of degree 1 in prices. So you can use Theorem
A2.6 and Euler's Theorem to get the result.
For point (b), find out from the Slutsky's equation Hicksian demand, and substitute into the
Hick's Third Law. Use logs for findings the demand.

1.66,

see the hint on page 524.

For part (a), minimimize p.x s.t u(x)<=u to get e(p,u)=p2.u - p2.p2/4.p1. Plug in y0=10
and p1=2,p2=2 to get u0=11/2.
For part (b), use part (a).
For part (c), mimic the proof on page 131 to show e(p,u)=k(u)e(p,1).

1.67

Since de(p,u)/dpi=xi(p,u),
dln e(p,u)/dln pi =de/dpi.pi/e
=xi.pi/e.

Capítulo 2

2.6

p1 p2 p1 p2 p1 p2 1 1
e=( )U V =( )e ( )y ( + )y
p 1+ p 2 ⇒ p 1+ p2 ⇒ p 1+ p 2 ⇒ p1 p2

By Roy’s identity:

−
∂V
( )
p2
y
x 1= ∂
∂V
p 1
=
p1
p 1+ p 2 x=
(
2
p )
p
y
2

∂y p 1+ p 2
Eq. 1) likewise Eq. 2)
From Equations 1) and 2) we will get Eq. 3)
p2 x 1
=
p1 x 2 √
y =√ x
p +p =
1
(2
p
p ) x2

1
1

2
y=
y
From equation 1) x1 x1 √x1 x2

p 1 + p 1 √ xx 12 = y
Use equation 3) again p1+ p 2 = √x 1x 2

( √ ) ¿ √ x1y x 2
−1
x1 p1= y
p 1 = 1+
x2 or ( √ x 1+ √ x 2 ) √ x 1 Eq. 4)

p 1 = p 1 √ xx 12 = y
( √ x 1 +√ x 2 ) √ x 1 Eq. 5)

2
Substitute these into V above ⇒ U=( √ x 1+ √ x 2 )

2.24.

(1) U(w)=-(b-w)**c  0
(2) U’(w) =c(b-w)**c-1 >0

(3) U’’(w) =-c(c-1)(b-w)**c-2 <0

Can we satisfy all three?

I. because of (1), if -1<c<1 and if c is even, then we need (b-w) >0

II. Because of (2), if |c-1| <1 and c-1 even, then we need (b-w) >0

C and (b-w)**c-1 need to have the same sign.

 if c is odd  c-1 is even  we need c>0

but if c is even  c-1 is odd  we need c<0 if (b-w)<0 and c>0 otherwise.

III. Because of (3), if c is odd  c-2 is odd  we need (c-1) and (b-w) to be of the same sign
since c already has to be positive because of (2).

But if c is even  c and (c-1) need to have the same sign. But when c is negative, b-w is
negative and vice versa. Therefore (c-1) and (b-w) will have the same sign.

R’(w) = (-2c)/(b+2cw)  increasing risk aversion only when b-w >0.

2.26

Since p1=p2=1, we know that y = x1+x2.

L = (lnx1+lnx2) + λ(y-x1-x2)

FOC: Find that x1 = y/3 and x2 = 2y/3

Find the indirect utility function to be u(y) = ln(2)+ 3ln(y)

u'(y) = 3/y

u''(y) = -3/y2

R(y) = 1/y > 0; Therefore, this function shows risk-averse behavior.

R'(y) = -1/y2 < 0; Therefore, this function displays decreasing absolute risk aversion.

2.4, 2.9, 2.14 and 2.19. I am not sure which differential equations are being referred to in
2.4 but assume they are (P.1) on page 83 together with e(p,u)=y.
2.4(a)We have de(p,u)/dp=x(p,e(p,u)) (a vector) and budget balancedness gives
p.x(p,y)=y. Then p.de(p,u)/dp=p.x=y=e(p,u). Now use Euler's Theorem.
2.4(b)If e(p,u) is homogenous of degree one in p then Theorem A2.6 implies de(p,u)/dp is
homogeneous of degree zero in p. Then de(tp,u)/dp=de(p,u)/dp so x(tp,e(tp,u))=x(p,e(p,u))
and hence x(tp,ty)=x(p,y) as e(tp,u)=te(p,u)=ty since e(p,u) is homogeneous of degree one
in p. (Actually, you have to tweak Theorem A2.6 a little.)
2.9(a)Homogeneity of degree zero gives minus p2.dx2/dp2=p1.dx2/dp1+y.dx2/dy. Budget
balancedness (BB) gives p1.x1+p2.x2=y so differentiating with respect to (wrt) p2 gives
p1.dx1/dp2=-x2-p2.dx2/dp2=-x2+p1.dx2/dp1+y.dx2/dy(*). Now differentiate (BB) wrt y
and multiply thru by x2 to get p1.x2.dx1/dy=x2-p2.x2.dx2/dy (**). Add (*) and (**), cancel
p1>0 and use (BB) again to get dx1/dp2+x2.dx1/dy =dx2/dp1+x1.dx2/dy. This is the
required condition.
2.9(b) Budget balancedness (BB) and WARP imply x(p,y) is homogeneous of degree zero in
(p,y) (page 88). So by part (a), with two goods, BB and WARP imply that the Slutsky matrix
associated with x(p,y) is symmetric. WARP and BB imply the Slutsky matrix is negative
semidefinite (pages 88-90). So two goods, BB and WARP imply symmetry, negative
semidefinite and BB. So by Theorem 2.6 x(p,y) is generated by utility maximizing behaviour.
This means that R has no intransitive cycles.
2.14 A={a1,a2,...,an}. Use G1 to make a series of pairwise comparisons and construct the
ascending series b1<=b2...<=bk<=...<=bn. Next use G2 to rule out the possibility of an
infinitely ascending chain, that is to rule out bn<=b1.
2.19 An individual is risk averse if and only if (iff) u(pw1+(1-p)w2)>pu(w1)+(1-p)u(w2)iff u
is strictly concave (pages 105 and 445).

Capítulo 3

Exercice 3.2

If f(X1X) is homogeneous of degree 1 implies by Euler Theorem that:

f=f1*x1+f2*x2

If we divide by x1, then

AP1=MP1+f2*X2/X1

Rearranging, (AP1-MP1)=f2*X2/X1 (1)

If we assume, X2 and X1 > 0, then if AP1 is increasing, implies that

DAP1/dx1= (MP1-AP1)*1/x1>0

And because x1>0 implies that MP1>AP1

Therefore, in (1), f2<0, and f2 is the MP2

QED

3.2

We know:

y = f(x1,x2) ; AP1 = y/x1 = f(x1,x2)/x1

Find dAP1/dx1:

(assume that "d" means the partial derivative sign)

dAP1 = df . 1 - f + df . dx2 - 1

dx1 dx1 x1 x12 dx2 dx1 x1

Therefore:

dAP1 = (1) . [ df . x2 . (-1 + dx2 . x1)]

dx1 x12 dx2 dx1 x2

If AP is increasing, then dAP1/dx1 must be positive.

We know that (dx2/dx1)*(x1/x2) is negative.

Therefore, (df/dx2), MP2, must be negative.

Exercise 3.3

Let Qi = Ax x x ..............x be a linearly homogenous production function.

Qi / x1 =  Ax x x .........x = MP1(x1)

Qi / x2 =  Ax x x ..........x = MP2(x2)

..............................................................

..............................................................

..............................................................

Qi / xn =  Ax x x .........x = MPn(xn)

Qi = ( Ax x x .........x ) x1 + ( Ax x x ..........x ) x2+......

( Ax x x .........x ) xn

= MP1(x1) x1 + MP2(x2) x2 +.............+ MPn(xn) xn

= MPi(xi) xi [i=1,2,3.........n]

Question 3.15(b)

Write y ( , x) = y (1, 2,…n, x1, x2,…,xn) = ( i xi )1/.

We have i = 1 and i > 0. Also xi  0.

Let xk = max {x1, x2,…,xn}. If there is more than one such maximum, choose
any one of them.

Then  i xi  k xk.

Also ixk  i xi so  ixk = xk  i xi.

Hence xk  i xi  k xk and so xk  (i xi)1/  k 1/ xk.

Since lim + k 1/ = 1, lim + (i xi)1/ = xk = max {x1, x2,…,xn}.

By matter of definition y ( , x) = 0 if  < 0 and some of the xi = 0.

For xi > 0, note that y - ( , x) = 1 / y ( , 1 / x).

But lim + y ( , 1 / x) = max {1/x1, 1/x2,…,1/xn} = 1 / min {x1, x2,…,xn} so

lim  -  y ( , x) = min {x1, x2,…,xn}.

3.31 in JR
Let si be the input share for input i. Show that, for any cost function,

si = dln[c(w,y)]
 dln(wi)

Verify by checking a Cobb-Douglas cost function.

Input share is defined as: si = wixi(w,y)

c(w,y)

C-D c(w,y) = Awiawjby

(1) Take the natural log: ln c(w,y) = (lnA)(alnwi)(blnwj)(lny)

Now we want to prove that a is equal to si

(2) Cond. Input Demands: xi(w,y) = dc(w,y) = aAwia-1w2by = ac(w,y)

dwi wi wi

(3) xi(w,y) = ac(w,y) [Mult. Both sides by wi ] wi xi(w,y) = ac(w,y)

wi

(4) [Div. both sides by c(w,y)]  wi xi(w,y) = a

c(w,y)

As was given in the original question, we proved a = si

(5) Take the partial of (1) with respect to wi  dlnc(w,y) = dalnwi

 dwi dwi

(6) After some algebraic manipulation, we get dlnc(w,y) = dalnwi

*** (7) Divide (6) by dlnwi  dlnc(w,y) = a = si ***

dlnw i

3.33

Calculate the cost function and the conditional input demands for the linear production
function, y=ni=1 ixi..

All xi are perfect substitutes.

FOC:  wi=xi

 Indeterminate: any combination of x’s will do.

However, corner solution if wi /xi=minw1/a1,, wn/an so that xi=y/ai .

3.40

Solution :

a) Solution: Cobb-Douglas function:

Min TC=w1x1+w2x2

S.T y=x1x2

L= w1x1+w2x2+( y-x1x2)

FOC:

=W1X1/y= W2X2/y
 x1=[(w2/w1) (/)y]1/+

x2=[(w1/w2)  (/)y]1/+

TC=W1/+W2/+y1/+

=[(/)/++(/)/+]

b)

CES function

Min TC=X1W1+X2W2

S.T y=A(X1e+x2e) 1/e

L = X1W1+ X2W2+ [Y-A(X1e+x2e) 1/e]

FOC :

L/ X1=W1-A1/e(X1e+x2e) (1/e)-1 eX1e-1=0

L/ X2=W1-A1/e(X1e+x2e) (1/e)-1 eX1e-1=0

w1=A(X1e+x2e) (1/e)-1 X1e-1

w2=A(X1e+x2e) (1/e)-1 X2e-1

w1/w2=(x1/x2) e-1

x1=(w1/w2)1/e-1 x2

x1=y/A[1+(w1/w2)e/e-1]-1/e
x2=y/A[1+(w1/w2)e/e-1]-1/e(w1/w2)1/e-1

TC=W1X1+W2X2

= y/A(W1 e/e-1+ W2 e/e-1) e/e-1

Let e/e-1=r

TC=AY(W1r+ W2 r)1/r
Exercise 3.52
a)Marginal cost= 2y(w1+w2)

Average cost=y(w1=w2)

Supply function: P=2y(w1+w2)

b) By Sheppard’s lemma we know that dC/dw1=x1=y**2

Then y=x1**0.5

Then , the marginal productivity of x1, MP1, is 0.5*(x1)**(-0.5)

The input demand function is obtained by equalizing P*MP1 with w1

 x1=P**2/(2*w1)**2

If w2 increases, MP1 is not affected by the change in w2. If output price is constant, then, the
input demand curves don’t move. Output will be smaller but this reduction will go together
with a reduction in x2, not in x1. If output price was going up because firms still want to
produce the same amount of y as before, then input demand curves would shift upwards. In
that case, if w1 stays unchanged, firms will hire more x1. The effect of the price increase on x2
on the other hand will depend on the elasticity of input demand function for x2 and on the
magnitude on the output price increase relative to the increase in w2. We will have a shift of
the x2 demand curve simultaneous with a movement along the curve. The total effect is not
predictable.

Exercise 3.53

A utility produces electricity to meet the demands of a city. The price it can charge for
electricity is fixed and it must meet all demand at that price. It turns out that the amount of
electricity demanded is always the same over every 24-hour period, but demand differs from
day (6:00 a.m. to 6:00 p.m.) to units are demanded. Total output for each 24-hour period is
thus always equal to 7 units. The utility produces electricity according to the production
function

Yi=(KFi)**(1/2), i=day, night,

where K is the size of the generating plant, and Fi is tons of fuel. The firm must build a single
plant; it cannot change plant size from day to night. If a unit of plant size costs w of k (wk) per
24-hour period and a ton of fuel costs w of f(wf), what size plant will the utility build?

min C=wkK+wf F1+wfF2 subject to y1=(KF1)1/2 , y2=(KF2)1/2

L=wkK+wfF1+wf F2 +1 (y1-(KF1)1/2)+2(y2-(KF2)1/2)

L/K=wk1/21K-1/2F11/21/22K-1/2F2 ½=0

L/F1=wf1/21K1/2F1-1/2=0  1=(2wfF11/2)/K1/2

L/F2=wf1/22K1/2F2-1/2=0  2=(2wfF21/2)/K1/2

plug 1 and 2 into L/K and simplify:

0=wk(wf/K)(F1+F2)  wk=(wf/K)(F1+F2)

or F1+F2 =(wk/wf )K

we know that y12=KF1 and y22=KF2  y12+y22=K(F1 +F2 ) 

F1 +F2 =(y12 +y22 )K plug it into wk

 wk=(wf /K2 )(y12 +y22 ) solve for K

K*=(wf /wk )1/2 (y12 +y22)1/2

Capítulo 4

JR Exercise # 4.6

Cj(q) = aq + bjq2

a) MC = a + 2bjq

P = MC

a + 2bjq = P

Q* =

bj  0  P 0 That will govern the amount produced by each firm. The firms will not
*
produce equal amounts as Q is also function of bj which is variable.

b) Q* =

If bj  0 Then P  a for all firms. This will imply losses as TR  MC.

Exercise 4.7

a)

2− p
Market Demand = β

J ( p−a )
Market Supply = 2b

Equating demand with supply, we obtain

¿ Ja β+2αb
p=
Jβ +2 b

¿ J ( α−a )
q=
Jβ+2b

b)

The monopolist in this case would equate MR = MC and the optimum number of firms in the
industry = 1, i.e. J= 1

α−a
( α −a
)
¿ ¿
q = m p m=α −β
In this case, 2 b+ β 2 b+β

c)

j
We have decreasing returns to scale in this case. In this case J ⇒ ∞ and q ≈0
This is because, as new industry attracted by positive profits, this will lower the market price
and will make the existing firms become smaller.
Exer 4.8 in JR
We have a Cournot Oligopoly where J=2. Let each have constant AC and
MC, but 0<= c1 < c2. Show firm 1 will have greater profits (Pr) and
produce a greater share of market output than firm 2 in a Nash
Equilibrium.

Price (P) = a-bQ where Q = q1+q2 and c1<c2

Pr1 = Pq1 – c1q1  (a-bQ)q1 –c1q1  [a-b(q1+q2)]q1 –c1q1

FOC: dPr1 = 0  q1 = (a-c1) - .5q2  This is Firm 1’s Reaction Function

dq1 2b

FOC: dPr2 = 0  q2 = (a-c2) - .5q1  This is Firm 1’s Reaction Function

dq2 2b

We know q1 = Q – q2 and q2 = Q – q1, so plugging the respective values

into the Reaction Functions, we get:

q1 = (a-c1) - .5Q - .5q1  q1 = (a-c1) - Q

2b b

q2 = (a-c2) - .5Q - .5q2  q2 = (a-c2) – Q

2b b
*** We are given that c1<c2, therefore (a-c1) < (a-c2) as given above. This
shows that q1>q2, meaning firm 1 produces a greater market share. ***

Pr1 = Pq1 – c1q1  q1(P-c1)

Pr2 = Pq2 – c2q2  q2(P-c2)

But we already know that q1>q2 AND c1<c2, so (P-c1) > (P-c2). Let us call

(P-c1) X and (P-c2) Y.

q1(X) > q2(Y) since q1>q2 and X>Y

*** Thus, Pr1 > Pr2 because Pr1 = q1(X) and Pr2 = q2(Y) ***

4.9

Stackleberg Duopoly

P=100−( q 1 +q 2 )
c 1=10q 1
c 2=q 22
a) Calculate the market price and each firm’s profit assuming that firm 1 is the leader and

Π 2=P⋅q 2−c 2
Π 2=( 100−( q 1 +q 2 ) )⋅q 2−10 q 22
2
Π 2 =100 q 2−2 q 2 −q 1 q 2
δΠ 2
δq 2 =100−q 1 −4 q 2=0
⇒ 4 q 2 =100−q 1
1
⇒ q 2 =25− q 1
4
firm 2 is the follower.
Π 1=P⋅q 1−c 1
Π 1=( 100−( q 1 +q 2 ) )⋅q 1−10 q 1
Π 1=90 q 1−q 1 −q 1 q 2
2

δΠ 1
δq 1 =90−q 2−2 q 1 =0
⇒ 2 q 1=90−q 2
1
⇒ q 1 =45− q 2
2

This gives both firms reaction functions.

If firm 1 is the leader, its profit function will look like this:

Π 1=90 q 1−q 12 −q 1 ( 100−q 1

4 )
Π 1=65 q 1 −3 q 1 2

4
δΠ 1
3
δq 1 =65− q 1 =0
2
⇒ q 1 =43 . 33

⇒ q 2=
100−43 .33
4
⇒q 2=14 . 17
⇒ P=100−( 43. 33+14 . 17 )
⇒ P=42 .5
∴ Π 1 =Pq 1 −10 q 1 =( 42 .5−10 )⋅43 .33=1408. 37
Π 2=Pq 2 −q 2 =42. 5(14 .17 )−(14 . 17)2 =401. 39
2
b) Do the same assuming firm 2 is the leader and firm 1 is the follower?

Π 2=100 q 2 −2 q 22 −q 2 (90−q
2
2
)
Π 2=55 q 2−3 q 2 2

2
δΠ 2
δq 2 =55−3 q 1=0
⇒ q 2 =18 .33

⇒ q 1=
90−18 .33
2
⇒q 2=35 .84
⇒ P=100−( 35 . 84+18 . 33 )
⇒ P=45 . 83
∴ Π 1 =Pq 1 −10 q 1 =( 45 . 83−10 )⋅35 .84=1284 . 15
Π 2=Pq 2 −q 2 =45.83(18 . 33)−(18 .33 )2=504 .08
2

c) Does firm 1 want to be leader or follower? Does firm 2 want to be leader or follower?

Firm 1 prefers to be the leader, because its profits are greater when it is the leader.

Firm 2 also prefers to be the leader, as its profits are also greater when it is the leader.

d) Both try to be the leader.

⇒q 1 =43. 33
q 2 =18.33
∴P=100−( 43 .33+18 .33 )=38. 34
Π 1=( 38 .34−10 )( 43 .33 ) =1227.97≺Π 1 F
F
Π 2=( 38. 34−18.33 )( 18 .33 )=366.78≺Π 2