Você está na página 1de 10

Economics 205A Fall 2012 K. Kletzer Problem Set 2 Sample Answers 1.

Ans: a) The command optimum solves

{
subject to

max
ct ,kt

et

c1 1 dt 1

kt 0
given initial k equal to k0 . b) The current-time Hamiltonian is

= f (k ) k c, k

for all t 0

H
The necessary conditions are

c1 1 + q (f (k ) (n + ) k c) . 1 c
=

q q =

= f (k) k c
kt

(k)

q,

q,

and
t

lim qt e

=0

and

k0 given.

c) The system you work with is

and

c c

= f (k)

= f (k ) k c. k
f (k

This has a steady state given by

) = +

and

= f (k ) k .

The solution path is given by the stable saddle path in your phase diagram. This converges to the steady state and is the optimum because it satisfies the transversality condition,

lim c s t
d) Linearizing, you get

et k t

= 0.

c =
and

c f (k

) (k k )

= k

f (k )

(k

k ) (c c ) = (k k ) (c c ) .

2 The eigenvalues are given by


=
1 2

4f

(k )

1/2

The root, , is negative, and the root + > > 0, so that we have saddl- path stability. The stable path is given by:
= c + (k0

where c0

k) ( ).

kt k = (k0 k ) et

and

ct

c = (c0 c) e

To first-order approximation, output is


yt

y = (y0 y) e

(from yt y

f (k

)( t

). Thus,

y t yt

(yt y yt

yt

so that at time t = 0, the growth rate of output is given by

which is increasing in y/y0 . The lower y0 relative to y, the higher is the growth rate of output. 2. Ans: a) The command optimum solves
{ct ,

= 1 y y0 y0
y 0

max

t ,nt ,kt

+1 } t=0

t u (ct ,

subject to

1 nt + t 0
given initial k equal to k0 .

kt+1 = f (kt , nt ) + (1 ) kt ct ,
and

nt , t , kt 0

for all t 0

b) The necessary conditions for an interior optimum are

u (ct , t ) t+1 , nt+1 ) t+1 , t+1 ) = 1 + f (k u (cc ct kt+1 t+1 f (kt , nt ) u (ct , t ) u (ct , t ) = for 1 > t > 0, nt ct t kt+1 = f (kt , nt ) + (1 ) kt ct , 1 = nt + t
and
t)

t u (ct , t ) lim ct kt+1 = 0 c, ) = , lim 0 u(c , These assume that limc 0 u( c t


t t t t t t t

and

f (k ,n ). We implicitly assume that f (k ,n ) and u (c , ) are strictly concave in their arguments and that f (k , 0) = f (0,n ) = 0. Corner solutions in and n are possible. The additional first-order conditions are
t t t t t t t t t t

= and that the Inada conditions hold for

k0 given.

f (k ,n ) u (c , n c
t t t t t

t)

c, u (
t t t t

t)

for

= 1,

=0

f (k ,n ) u (c , ) u (c , ) for n = 0, = 1. n c Now, let f (k,n) = Akn1 for 0 < < 1 and A > 0 constants. Let u (c, ) = log c + log . These will lead to solutions such that nt > 0 and t > 0, so that nt+ t = 1 for all t. The necessary conditions
t t t t t t t t t

and

become

ct =
1

1 n1 1 + Akt+1 t+1

k +1 c +1 c +1 c = 1 + A n +1 ct = (1 ) A kt , n 1 = 1 (1 ) Akt t c nt t t t 1 kt+1 = Akt nt + (1 ) kt ct, 1 = nt + t


t t t t t

and

c) Begin with the three conditions

t kt+1 = 0 lim t c t

and

k0 given.

and

ct+1 = 1 + A k +1 1 , ct n +1 c = (1 ) A k 1n n kt 1 k c . kt+1 kt = A n t
t t t t t t t t

The steady state is given by

and

These simplify to

k 1 = 1 1 + = + A n c = (1 ) A k 1 n n k 1 . c = A k n
1 k = A 1 , n + 1 c = A 1 (1 ) 1n + 1

and

+ . = k
c

1 1 n = 1 + 1 + which is strictly between zero and one for > 0 and 1 > > 0.

Y ou might simplify further to find that n is given by

We can reduce the system of three equations to a dynamical system in c and k by eliminating n. However,

lets linearize all three and simplify our system from there. Linearizing the Euler condition about the steady state leads to the equation

k dk +1 dn +1 dk +1 dn +1 (1 ) A n n = (1 ) ( + ) k n k where the notation dx x x . Next, linearize the first-order condition for consumption of goods and
t t

dc +1 dc c c

leisure. This is
dct c

Lastly, linearize the equation of motion for k as k 1

dk dn dn 1 = . n k n
t t t

dkt+1 dkt

= A

dkt + (1 ) A

= dk + A
t

(1

Now, we need to eliminate dkt+1 and dnt+1 from the linearized Euler condition so that we have dkt and
dnt

k n

dn dk dc
t t

dnt

dct .

on the right-hand side of the condition. To do this, use the first-order condition for goods-leisure choice

to write

1 dn = dk dc . + n k c dc dn dc 1 dc dk + = + t 1 n c 1n c k c 1 + n Then use dc +1 = (dc +1 dc )+ dc = c +1 + dc and dk +1 = k +1 + dk


1 1 n
t t t t t t t t t t t t t t t

to substitute into the Euler

condition to get (in steps)


t

c +1 = (1 )( + ) c

k +1 n +1 dn = (1 )( + ) dk (1 )( + ) k n k n dc dn 1 ( + ) c +1 + n +1 = 1 ( + ) + c 1n c 1 n 1 dk 1 n 1 n dc 1 n c +1 + k +1 + ( + ) 1 + + = 1 n n c k n c k
t t t t t t t t t t t t

dkt+1 k

+1 dn n
t

We replace

k +1 using the equation of motion for k and rearrange to get c +1 1 + (1 ) ( + ) 1 n


t

) ( + ) = (1 1 + 1n
n

n 1 + 1 n

1 n

c k

dct c

+ 1 dk + k
t

( + ) dn n

(1 ) ( + ) c +1 = 1 + ( + (1 ) ( + )) 1n
t

Now, we can eliminate dnt to get

where

c k

and

1n

= 1 + .
1

1 n

1 ( + ) 1 n 1 + 1 n

(1 ) ( + ) c c dkt k dc k 1 + 1n

t

We can also write the equation of motion for k eliminating dnt (which was done in the course of finding the linearized

ct+1 equation). This is 1n n kt+1 = + (1 ) ( + ) 1 + 1n

dkt
1

1+

( + )

1+

1n

1 n n

k c

dct

We can use the first-order condition for the goods-leisure choice to write

n +1 =
t

1 n n n 1 + 1
n
t

Y ou should note that the linearized

n +1 equation is a linear combination of the linearized equations of

+ A

c +1 k +1 k c
t t

motion for k and c. Therefore, you do not want to calculate the eigenvalues and eigenvectors for a threedimensional system (the determinant would be zero). Instead, you use the two-dimensional, non-degenerate dynamics. The system is given by

(1 ) ( + ) c +1 = 1 + ( + (1 ) ( + )) 1n
t

1+ n and the dynamics for n are given recursively from those for k and c.
t

k +1 =

1+

( + )

1 n

1n

1 ( + ) 1 n 1 + 1 n

(1 ) ( + ) c c dk k dc k 1 + 1n

t

1n c

dct

n + (1 ) ( + ) 1 + dk 1n n
t t

1n

Because the equation of motion for n expresses


t

n +1 as a linear combination of k +1 and c +1, we


t

have a two-dimensional dynamical system. This can be written as

where m12 and m21 are negative and m22 is positive. non-negative is

m11 has an ambiguous sign. It will be negative for small values of but not for sufficiently large values of . A restriction such that this matrix element is
1

c +1 m11 m12 k +1 = m21 m22


t

c c k k
t t

(1

(1

) ( + ) 0. 1 + 1 +
(1
)

Assuming this inequality holds, the determinant of the matrix is negative, so the dynamical system is saddlepath stable. means that the additional output at t will provide for an increase in solution for ct+s and kt+s for s d) The economy is initially in the steady state. Productivity, A, changes by dA for only period t. That

1 must be on the saddle path that converges to the original steady state, c and k. Labor supply will adjust along with c and c +s .
t t

c + k +1.
t t

Begining in

t + 1, the

Start with the effect of the productivity shock on the equation of motion where economy starts in the steady state:

kt =

6 0 because the

d (kt+1 + ct) = kn1dA + (1 ) Akndnt. We use the linearized first-order condition for the goods-leisure choice to find dnt , dct = dk dn dn = dn dn c k n 1 n n 1 n where with dk = k k = 0 in the steady state. This gives us 1n c dn = 1n + 1 dc
t t t t t t t t

which substitutes into the equation of motion to get

d (k +1 + c )
t t

A kdA (1 ) A A + + A kdA 1 t n + 1 n + 1
dc .

1 A 1 + 1 n + 1 dc (1 ) n

after substituting expressions for

k n

n and 1 from part c. c

This gives us one of the relationships we need,


dkt

+1 +

n 1 n n 1 n

+ 2 dc = A kdA. + +1
t

Another is the equation of the stable eigenvector which as the positive slope,

Now, you can see we need the Euler condition to get a relationship between dct+1 and dct. Use one of the earlier versions that has

dc +1 m22 dk +1 = m21 .
t

c +1
t

(1 ) ( + )
n 1 + 1 n n

k +1 in it,
t

) ( + ) = (1 1 + 1n
where the last equation uses dkt

(1 ) ( + ) 1 n 1 n n 1 + 1 n (1 ) ( + ) 1 n 1 n n 1 + 1 n

1 n

dct

c c +k +k dk k +1
t t

1 n

= 0.

dct

c +k dk +1
t

The three conditions solve for dkt+1 and dct simultaneously from

dkt+1 +
and

n 1 n n 1 n

+ 2 dc = A kdA, + +1
t

m22 + (1 ) ( + ) = m 1n
21

n ) 1 n 1 (1 ) (1+ n
1 + n
1+

(1 ) ( + ) 1 n n 1 + 1 n
1

dct
1

(1 ) ( + ) 1 n n 1 + 1 n

c dk , t+1

7 which imply that

we have

The shock increases consumption of both goods and leisure (that is, labor supply, nt, falls) at time t, raises investment (dkt+1 ) at time t. Consumption and leisure remain higher than in the steady state for t but converge back to their steady-state values. steady-state level from kt+1 . 3. Ans: a) The two linearized equations are correctly given in the text as
t

dc > 0 and dk +1 > 0, dA dA n dn = 1 c < 0. n dc 1 n + 1


t t t t

k +2 is negative as the capital stock declines back to its

+1

(1 ) dq f (k ) dk = q +1 + f (k )k +1 k dq = k +1 ,
t

where we use the notation, dqt

q q , as before.
t

All we do is substitute the second into the first to get

q +1 = f (k ) k
t

dqt

f (k ) dk

and

for the linearized system. b) This was redundant - apologies. c) The system is

k +1 = k dq
t

which has a negative determinant, k f

q +1 = (k f k +1
t

(k ) k )

f
2

(k )

(k ).

dqt dkt

The eigenvalues solve

1 2

f (k ) k

(k ) k

One solution is < 0 and the other is + > f


=

4k f

(k )

1/2

1 The slope of the saddle path that converges to the steady state is negative, and the slope of the divergent
saddle path is positive. Note that the locus,
dq q = 0, dk = ff
t+1
t t

(k ) k

> 0. The eigenvectors satisfy

k + = 0, is has zero slope and the locus, q


t

t+1

= 0, as negative slope.

The

slope of the (linearized) stable saddle path is between these. This can be seen by comparing the slope of
k
(k ) (k )k

, , to the slope of the saddle path, k

f (k ) k

f (k

<

1 2

(k ) k

(k ) k

1/2 k 4 f (k )

Rearranging and squaring both sides (remember to reverse the inequality symbol since each side is a negative

8 number) gives us
k f (k ) 2 f (k) k

f 4 f

(k ) k

f f

(k ) k

(k ) k

k f (k ) +4 f (k) k

k f (k) f (k ) k

(k ) k

k 4 f (k ) + 4

k f (k ) f (k ) k

>

f (k ) k 2 4 k f (k ) .

d) Let k0 be less than k , the steady-state capital stock. How do investment and q behave over time? This is limt tqtkt+1

= 0. The is given by a pair (k0, q0) on the stable saddle path. Thus, q0 q = (k k0 ) k where q = 1+ and k by f (k ) = + + + 2 . Both kt and qt converge toward their steady-state
4. Ans: a) The problem is
V (kt ) =
{ct ,kt+1 }

For k0 < k , the solution for q0 is given by the requirement that the transversality condition is satisfied.

values at the proportionate rate . As k grows, q decreases.

max

c1 t

1 + V (k ) +1
t t t

subject to for ct

0 and k +1 0.
t

k +1 = (1 + A ) k c
t

The necessary conditions include the first-order conditions,


c t = t V (kt+1) = t+1,

and the envelope condition,

V (kt) = (1 + A ) t
which also can be written

V (kt+1) = (1 + A ) t+1. The necessary conditions include the equation of motion, kt+1 = (1 + A ) kt ct , the constraints, ct 0 and kt+1 0 and a transversality condition, limT t+T T kt+T 0. b) First, eliminate the derivative of the value function, V (kt ) and V (kt+1 ), to write the Euler condition

in terms of consumption as
. ct = (1 + A ) ct+1

1 1 s t ct+1 = ( (1 + A )) cs = ( (1 + A )) ct ct and ts T 1 1 (1 + A ) kt = cs + lim kT +1. T 1 + A 1+A s =t Imposing the transversality condition and non-negativity constraint for kT +1 , we get ts s t 1 s t 1 1 (1 + A ) kt = cs = ( (1 + A )) 1+A 1+A s=t s=t
( ) (

9 Next, solve the Euler condition backward and the equation of motion for the capital stock forward to get

ct

which leads to
(1 +

A ) k
1

The solution for ct is


ct

(1 + A )
1 1 1

ct.

(1 + A ) k . = 1 (1 + A )
t

c) First, calculate

= s=t 1 1 (st) ct s t ( (1 + A )) = 1 s=t = 1 1 1 1 ct . = 1 1 1 1 (1 + A ) 11 Next, use the solution for ct in terms of kt to substitute for ct in max Ut : 1 1 1 1 1 (1 + A ) 1 1 1 kt max Ut = (1 + A ) 1 1 1 1 1 1 (1 + A ) 1 1 1 kt 1 1 = (1 + A ) . 1 1 1
s t

max Ut =

st

1 c1 s 1

st ct

( (1 +

A )) st 1 1 1 st 1 s t
1
)

The value function is given by

V (kt ) = max Ut =
Note that this function has the form,

(1 +

1 1 A )

1 kt 1 1 . 1 1 1

V (kt ) = C
for constants C and D.
1 kt 1

1 kt 1

+D
k1 C t+1

d) Substitute your function into the dynamic programming problem,

C
subject to for ct

+ D = max
t

{ct ,kt+1 }

c1 t

1 + 1
t

0 and k +1 0.
t

k +1 = (1 + A ) k c

10 The first-order conditions are


Ckt = t (1 + A ) , Ckt +1 = t c t = t .

and

These lead to the equations,


1 1 1 1 kt+1 = (1 + A ) k and c +1 = (1 + A ) c , which imply that c = Zk for some constant Z . Substituting into the integrated resource identity, s 1 (1 + A ) k = cs , 1 + A s=
t t t t t t t t

we verify the solution,

ct = 1 (1 + A )
1

(1 +

A ) k .
t

Você também pode gostar