Economics 202A Suggested Solutions to Problem Set 2

David RomerGalina Hale Spring 1999

1 Romer, 1.11. Embodied technological progress

a We modify rst the Solow model so that Y t = AtK tLt1, and the growth rate of A is now . We now have to establish the convergence to the balanced growth path and nd the growth rates of K and Y on the balanced growth path. Y Method 1: Using the hint, we dene k = AK L ; y = A L , where = 1, . Then

L1, K = K L = k y = A K1, = 2 A L A 1, , L A 1 _ = sY , K . Using this fact and taking logs We know that as before K and derivatives of the denition of k with respect to time, we get

diagram that the economy actually converges to the balanced growth path 1

_ _ L _ sAK L1, _ K A k , , , n = k = K , A , L = K L1, , , , n = 1s = sA 1 , K k , , n + + ; _ = sk , n + + k: k So, on the balanced growth path sk = n + + . Now we show using the

see Figure 1. Our function satises all the properties to insure that. If _ 0 and k converges to k. If k k, then k _ 0 and k also k k, then k converges to k.
c . c=0 . new c=0

c*

E new

new

. new k=0 . k=0

c*

B E A k* k* new

Figure 1: Convergence to the BGP. From the denitions of k and y we know that when k is constant, on the balanced growth path, y is constant and K and Y grow at the same rate of + n. Method 2: we can calculate growth rates without using k . _ _ _ _ Y A K L = + + 1 , Y A K L; gY = + gK + 1 , n: 2

Y must be constant, which implies that For gK to be constant the ratio K gY = gK on the balanced growth path. So we can derive that

Then, if the balanced growth path exists, the growth rates gY ; gK should be constant. Wee also know form the equation of motion of capital that _ K Y = g K = s , : K K

+ n = + n: gY = gK = 1 ,

_t = sAT Y t , J t. We will use b Now Y t = J tLt1, and J method 1 to show the convergence to the balanced growth path. J = J; j = J J = 1 A A L A 1, L 1, = j ; y = J 1L L A,

we see that our production function satises all the important assumptions. Taking logs and derivatives of the denition of j we nd that _ L _ _ j J_ , 1 A = , j J 1,A L = 1 , n; = j 1s , , , 1, _ = sj , 1 + n + j: j 1, By analogy with part a we can conclude that the economy converges to its balanced growth path determined by the following condition: 1 + n + j = 1 + + n + j: sj = 1 , On the balanced growth path, j and y are constant and thus J grows at the constant rate of 1 + + n and Y grows at the constant rate + n. 3

c We can determine from the result in b that

s j = 1 + + n +
"
1

and that

y = j =

"

which is the same as in Solow model except for the technological growth rate term. Thus, as in Solow model, the elasticity of y with respect to s will be . equal to 1, d As we did in section, consider the rst order approximation of y _ around _: the BGP y = y noting that since y = j , y _ = j ,1j
y _ = @y _ j y , y; @y y=y y _ = j ,1sj , n + + 1 + j = = sj 2,1 , j n + + 1 + = 2,1 = sy , yn + + 1 + ;

s 1 + + n +

1,

,1 @y _ j y , y = 2 , 1sy , n + + 1 + = y=y @y = 2 , 1n + + 1 + , n + + 1 + = = ,1 , n + + 1 + ;

y _ =

,1 , n + + 1 + y , y:

If we assume that = g, then this speed of convergence is higher than in is positive. Solow model, because = 1,

Natural resources

In this version of Solow model Y = F Z; R is CRS. 4

" =,

1 = " = , 1 : "

Z @ ln F FR Z @ ln R

a The factor shares are dened as usual:

R = FR R FZ Z ; Z = : FZ Z + FR R FZ Z + FRR

Z = FZ Z and ln Z = ln FZ + ln Z . The elasticity is then Then FR R R FR R R

" =

FZ Z @ ln F @ ln R = R +1="+1=1, 1 Z Z @ ln R @ ln R @ ln Z R = 1, 1 Z @ ln R

Z = 0:02, therefore we can write that Z 50 = b We are given that ln_ R R Z Z : Z R e = e , or taking logs, ln R = 1. R 0, which is equivalent to Z R We can rewrite the result in a in discrete terms now: ln Z = 1 , 1 ln Z :
0 02 50 50 0 1

We are given that = 0:8, therefore 1 , = ,0:25. Also, because R = 0:15 : Z we can calculate ln R 0 = ln : = ln 5:6 = 1:72. Substituting, we get that
1 0 85 0 15

Z Thus, R = 4:36 and, solving for R, we get that R = 0:3 = 30. c i = 0:6, therefore 1 , = ,0:66
1

Z = , 0 :25 = ln Z 50 = 1:72 , 0:25 = 1:47: ln

R R

Z 50 = 1:72 , 0:66 = 1:06 = R ln = 0:53 = 53:

R

1 = 0:2 ii = 1:25, therefore 1 ,

Z ln 50 = 1:72 + 0:2 = 1:92 = R = 0:17 = 17: R

As the elasticity of substitution between factors increases, which means that scarce factor is easier to substitute for, the share of this factor grows slower and slower. This is the result we would expect to see.

3 Romer, 2.3. Log utility

We have to nd the path of consumption per worker in the Ramsey model given that uC = ln C . The Household is maximizing 2.1 in the book subject to 2.5 in the book. We will normalize the number of households to 1 for simplicity. We also know that the budget constraint is binding for the maximizing household. max s.t.
Z 1
0

Z 1
0

e,t ln C tLtdt
Z 1
0

e,R t C tLtdt = K 0 + Where W is wealth plus the present value of the labor income. Then we have to nd C W; Rt; ; initial conditions. As this problem is the special case of the one considered in class with = 1, we can use Lagrangian to solve it. I will use Hamiltonian to show you the method. But rst we have to rewrite the life-time budget constraint as a period budget constraint. We can do this simply by dierentiating. First, rewrite budget constraint as of time s:
Z1

W | , R t e AtwtLtdt;

s
Z1

e, R t ,R s C tLtdt = K s +

Z1

e, R t ,R s AtwtLtdt:

K ; c = C and divide through by AsLs: Now dene k = AL A

Z1 , Rt,Rs+n+g t,s e ctdt = ks + s e,Rt,Rs+n+gt,swtdt: s

We will dierentiate this with respect to s, but notice rst that @ Rt , Rs = ,rs; from the denition of Rt, and thus the result of the dierentiation is recall Leibniz Rule:

@s

,cs + s rs , n , ge, R t ,R s Z1 _ ks , ws + s rs , n , ge, R t ,R s

Z1

+n+g t

,sctdt = ,swtdt:

+n+g t

We can now take the constants out of the integral note that rs is constant with respect to t and replace the remaining integrals with ks, because that is what they are. We get k_s = ws , cs + rsks , n + gks; which is the period constraint we are going to use in our maximization. But before writing down the Hamiltonian we have to rewrite the objective function in terms of per unit of eective labor". Note that ln C t = ln ct+ ln At = ln ct + ln A0egt = ln ct + ln A0 + gt. I will also divide the maximand by the constant L0. max ct

Z1
0

e, ,n tln ct + ln A0 + gtdt

s.t.

The present value Hamiltonian is then

k_t = wt , ct + rtkt , n + gkt:

H = e, ,n tln ct + ln A0 + gt + wt , ct + rtkt , n + gkt

and the rst order conditions are @H = 0 = 1 = ; for all t; @c ct @H _ , , n = ,rt , n , g = _ = + g , rt: , @k = 7

If we dierentiate the rst one and plug it in the second one, we get c_t = rt , , g ct which implies that ct = c0eR t , g t and gives us the path of consumption per unit of eective labor. But we are interested in the consumption per person:
+

C t = ctAt = Atc0eR t ,

+g t

= A0c0eR t ,

+g

,gt = C 0eRt,t :

The only thing we are left to determine is C 0. We will use the life-time budget constraint to nd it. Rewrite the budget constraint substituting the expression for C t:
Z1
0

e,R t

+Rt

, ,ntC 0L0dt = W:

Note that the solution exists for n. Taking this integral we nd 1 = W = C 0 = W , n: C 0L0 , n L0 Finally, we can substitute it into our equation for C t to get
R t ,t C t = LW : 0 , ne

We see that consumption increases if the interest rate is higher than the intertemporal discount rate and its level is larger the larger the lifetime wealth and the smaller is the initial size of the family.

4 Romer, 2.6. Playing with the phase diagram

We will use our equations for the dynamics of k and c: c _ = c f k , , g
0

_ k

f k

, n + gk , c:

_ equation is unaected and thus the locus k _ =0 a When goes up, the k is unaected. In the equation for c _ the increase in will decrease the RHS, so for the c_ to be zero, f k must increase for every c which means that k must decrease for every c. Therefore the only change to the graph is the shift of c _ = 0 locus to the left see Figure 2. The new steady state values of k and c are lower then initial.
0

. new c=0

. c=0

c* c* new E new E . k=0

k* new

k*

Figure 2: goes up _ = 0 locus shifts b When the production function shifts downward, the k down, because for k to be constant, c must go down for every k. Also, for a given k, f k is now lower and therefore the c _ = 0 locus shifts to the left see Figure 3. As a result, the new steady state values of k and c are lower then initial.
0

t=0

time

Figure 3: Production function shifts down OR Rate of depreciation is positive c Now the depreciation rate is positive. This increases the break-even investment and the equations of motion change. In the equation of motion _ equation in the same manner as for c, is not part of , but it enters the k n + g. The new equation of motion will be: = _ = k
0

c _

f k , , g , c f k , n + g + k , c:
0

_ = 0 now, c must fall for every k, therefore k _ = 0 locus shifts down. To keep k To keep c _ = 0, f k must increase for every c, therefore k must fall and c _=0 locus shifts to the left see Figure 3.

10