Or Solutions

FOUNDA TIONS OF INTERNA TIONAL
SOLUTIONS MANUAL
Maurice Obstfeld, Kenneth Rogo, and Gita Gopinath
Version 1.1: February 27, 1998
MACROECONOMICS
c MIT Press, 1998
This manual contains the solutions to the end-of-chapter exercises in Foundations of International Macroeconomics by Maurice Obstfeld and Kenneth Rogo. All solutions are also available at
http://www.princeton.edu/ObstfeldRogoBook.html
Along with the solutions, the website contains overheads for use in teaching Foundations of International Macroeconomics, as well as other information and accessories, such as typos in the rst (Fall 1996) printing, new exercises (when available), and beta versions of new chapters (if and when available).
Obstfeld and Rogo are grateful for assistance and feedback from students in their graduate international nance and macroeconomics courses at Berkeley and Princeton. The authors wish to thank especially Giovanni Olivei and Stefan Palmqvist for their help in preparing this solutions manual.
Foundations of International Macroeconomics1
Workbook2
Chapter 1 Solutions
1. (a) The intertemporal budget constraint can be expressed as C2 = (1 + r) (Y1 C1 ) + Y2 . Substitute this expression for C2 into lifetime utility U(C1 , C2 ) to obtain U = U [C1 , (1 + r) (Y1 C1 ) + Y2 ] . (1)
Taking the total derivative of U with respect to C1 and equating it to zero, one gets the following Euler equation for optimal consumption: U(C1 , C2 ) U(C1 , C2 ) = (1 + r) . C1 C2 (b) Total dierentiation of expression (1) with respect to r gives dU dr U(C1 , C2 ) dC1 U(C1 , C2 ) dC1 = + (Y1 C1 ) (1 + r) C1 dr C2 dr " # U (C1 , C2 ) U(C1 , C2 ) dC1 U(C1 , C2 ) = (1 + r) (Y1 C1 ) + C1 C2 dr C2 U(C1 , C2 ) = (Y1 C1 ), (2) C2
" #
where the Euler equation from part a gives the last equality. (Notice another instance of the envelope theorem.)
By Maurice Obstfeld (University of California, Berkeley) and Kenneth Rogo (Princec ton University). MIT Press, 1996. 2 c MIT Press, 1998. Version 1.1, February 27, 1998. For online updates and corrections, see http://www.princeton.edu/ObstfeldRogoBook.html
1
(c) Recall the relationship derived in part b, dU (C1 , C2 ) U (C1 , C2 ) (Y1 C1 ) . = dr C2 Because the marginal utility of consumption is positive, the country will benet from an increase in r if and only if Y1 C1 > 0, that is, if it is a net lender in period 1. In that case, the increase in r corresponds to an improvement in the countrys intertemporal terms of trade. (d) Write expression (1) as U = U [C1 , (1 + r) (W1 C1 )] . Dierentiation with respect to W1 gives dU U (C1 , C2 ) = (1 + r) . dW1 C2 (By the envelope theorem, the two terms in dC1 /dW1 cancel each other.) If b b dW1 = r(Y1 C1 ), where r dr/(1 + r), then by the Euler equation, dU = (1 + r) But eq. (2) in part b implies that the welfare eect of a percentage gross b interest rate change r is the same, dU =
b Thus the interest-rate change alters lifetime wealth by the amount r (Y1 C1 ) . This is the usual type of formula for a terms-of-trade eect, as computed in static international trade theory.
U(C1 , C2 ) U (C1 , C2 ) b b r(Y1 C1 ) = r (Y1 C1 ) . C2 C1
U(C1 , C2 ) U(C1 , C2 ) b (Y1 C1 )dr = r (Y1 C1 ) . C2 C1
2. (a) Substitution of the consumption Euler equation C2 = (1 + r)C1 2
into the intertemporal budget constraint gives 1 Y2 C1 (r) = Y1 + . 1+ 1+r (b) Home saving is derived as S1 (r) = Y1 C1 (r) = 1 Y1 Y2 . 1+ (1 + )(1 + r) (3)

A parallel expression gives Foreign saving. (c) Global equilibrium requires that
S1 (r) + S1 (r) = 0.
Substituting into this relationship the expressions for saving from part b gives Y2 Y2 + 1 + 1 + 1+r = . Y1 Y1 + 1 + 1 +
(d) In autarky, S1 (rA ) = S1 (rA ) = 0, so that
1 + rA = 1 + rA =
Y2 , Y1 Y2 . Y1
Using these solutions to eliminate Y2 and Y2 from the expression for the equilibrium world interest rate in part c, one obtains r= (1 + )Y1 (1 + )Y1 rA + rA . (1 + )Y1 + (1 + )Y1 (1 + )Y1 + (1 + )Y1
Because the world interest rate is a weighted average of the two autarky rates, it must lie between them.
(e) The expression for Homes current account is given by eq. (3) and can be written as Y2 CA1 = S1 = 1+
1 1 1 + rA 1 + r
r rA Y2 = , 1 + (1 + rA )(1 + r)
"
where the autarky interest rate formula of part d has been used to make the substitution Y1 = Y2 /(1 + rA ) in eq. (3). Plainly Home will run a date 1 current account surplus, CA1 > 0, if and only if rA < r. An autarky interest rate above r implies CA1 < 0. (f) From the expression for the equilibrium world interest rate in part d, it is easy to see that an increase in Y2 /Y1 raises r by raising Foreigns autarky interest rate. Moreover, given that U1 (C1 , C2 ) dU1 (C1 , C2 ) = (Y1 C1 ) , dr C2 (see part b of question 1), the consumption Euler equation and the expression for C1 (r) from part a imply. 1 1 dU1 1+ = Y1 Y2 dr 1 + r Y1 + Y2 /(1 + r) 1 + (1 + )(1 + r) " # (1 + r)Y1 Y2 1 = 1 + r (1 + r)Y1 + Y2 Y2 (1 + r) Y1 = Y2 1+r (1 + r) + Y1 " # r rA = . 1 + r (1 + r) + (1 + rA )
" #" #
If the world interest rate exceeds Homes autarky rate, Home runs a date 1 current account surplus, as shown in part e. In that case a higher Foreign rate of output growth causes r to be higher, and Homes welfare is enhanced via the favorable intertemporal terms of trade eect. Homes welfare rises 4
because the rise in Foreign growth widens the spread between autarky interest rates, increasing the gains from intertemporal trade. If Home were instead a date 1 borrower, a rise in Foreign output growth would reduce the gains from trade and also reduce Home welfare. 3. (a) From the equality of the interest rate r and the marginal product of capital 1 A2 K2 = r, one obtains A2 1 K2 = . r (b) From the identity I1 = K2 K1 , substitution of the above expression for K2 gives the function I1 (r). (c) Given that utility is logarithmic, consumption at time 1 is a fraction 1/1 + of lifetime wealth. From the intertemporal budget constraint C1 + I1 + one readily gets C1 (r). (d) Given I1 (r) from part b, C1 (r) from part c, and noting that I2 = K2 , one can rewrite C1 (r) as 1 1 K1 + Y1 + C1 (r) = 1+ 1+r r
"
1
C2 + I2 Y2 = Y1 + , 1+r 1+r
A2
1 1
It is then straightforward to compute saving as S1 (r) = Y1 C1 (r), and check 0 that S1 (r) > 0: S1 (r) = r
0
A2
1 1
1 +r > 0. 1 + r(1 + r)2
"
Refer to sections 1.3.2.2 and 1.3.4 in the book for analysis of the slope of the saving schedule in the case of a general utility function. In this particular 5
exercise where u(C) = log(C), the intertemporal substitution elasticity is equal to 1. Thus the substitution and income eects cancel each other and only the wealth eect ( which raises saving) remains. The slope of the saving schedule therefore is unambiguously signed. 4. (a) Recalling eq. (25) of Chapter 1 in the book, C2 = (1 + r) C1 , one sees that C1 = C2 as goes to 0. (b) The expression for C1 (r) can be obtained using the intertemporal budget constraint together with C1 = C2 . In a two-period model CA2 = Y2 + r (Y1 C1 ) C2 = (Y1 C1 ) = CA1 , one can verify that C2 (r) = C1 (r) by substituting the expression for C1 (r) into the second of the preceding equalities and solving for C2 (r). (c) Dierentiating the expression for C1 (r) obtained in part b, one has dC1 Y1 Y2 = . dr (2 + r)2 When = 0 the pure substitution eect of an increase in r is equal to zero, so that the only eect is the intertemporal terms of trade eect. Given that Y1 Y2 , 2+r the country will run a current account surplus if and only if Y1 > Y2 . In that case an increase in r represents a terms of trade improvement, and consequently C1 increases. S1 = Y1 C1 = (d) Under the postulated utility function the intertemporal Euler condition is 1/ 1/ C1 = (1 + r) 1/ C2 , 6
or C2 = (1 + r) C1 . As 0, the Euler equation approaches C2 = C1 . 5. The global equilibrium condition is

Y1 + Y1 = C1 + C1 .
Assume that Y1 changes. In the notation of section 1.3.4, with R = 1/(1 + r),total dierentiation of the global equilibrium condition gives
C1 (R, W1 ) dW1 dC1 (R, W1 ) dC1 (R, W1 ) dR 1= + , + W1 dY1 dR dR dY1
"
from which
C1 (R, W1 ) dW1 1 dR W1 dY1 # =" dY1 dC1 (R, W1 ) dC1 (R, W1 ) + dR dR
follows. Since dW1 /dY1 = 1 and C1 (R, W1 )/W1 < 1 if both periods consumptions are normal goods, a rise in Y1 lowers the world interest rate r (i.e., dR/dY1 > 0) provided a fall in r (rise in R) raises total world consumption. As long as the condition for stability (in the Walrasian sense) of the world market equilibrium is satised, however, the last condition holds. The eect of a rise in Y1 is symmetric, whereas that of a rise in Y2 follows from 1 C1 (R, W1 ) dR 1+r W1 #. =" dY2 dC1 (R, W1 ) dC1 (R, W1 ) + dR dR 6. (a) Date 1 equilibrium (with internationally identical isoelastic preferences) is 1 A2 F (I1 ) + I1 0 = Y1 Y1 I1 + I1 1 + (1 + r)1 1+r 7
" #
(4)
Y1
1 A F (I1 ) + I1 Y1 I1 + 2 I1 , 1 + (1 + r)1 1+r
"
where r = A2 F 0 (I1 ) determines investment in Home (with a similar equation in Foreign) and where we have substituted I2 = I1 , and similarly for Foreign. (We assume K1 = 0 and depreciation = 0.). We rst dierentiate to nd the equilibrium change in the interest rate, dr. Using the equation for dC1 /dr on p. 29 of Chapter 1 and the envelope theorem, we see that the dierential of the equilibrium condition above is 0 =
CA1 C2 /(1 + r) CA1 + C2 /(1 + r) dr + dr 1 + r + (C2 /C1 ) 1 + r + (C2 /C1 ) F (I1 )/(1 + r) dA2 1 + (1 + r)1 ! I1 I1 I1 + dr dA2 . r r A2
Solving for dr (and using the Euler equation to eliminate C2 /C1 ) gives F (I1 ) + [1 + r + (1 + r) ] dr = I1 A2
I1 I1 (C2 + C2 ) [1 + r + (1 + r) ] + 1+r r r
! dA2
> 0.
(5)
The eect on Foreigns current account CA = Y1 C1 I1 is 1
dCA = 1
CA C2 /(1 + r) I 1 dr 1 dr. 1 + r + (1 + r) r
Notice that I1 /r < 0. Accordingly, if CA < 0 (implying that Home has a 1 date 1 surplus, CA1 > 0), Foreigns date 1 current account improves whereas Homes worsens. For Foreign, the terms of trade and substitution eects of the rise in r both work to raise saving in this case, and since investment I1 must fall, CA rises and CA1 falls. 1
(b) To see what happens to when A rises, note that the positions of Home 2 and Foreign are simply reversed, but since CA1 > 0, the rise in the interest 8
rate has a positive terms of trade eect on Home that tends to increase C1 and reduce S1 . So we need (at the least) to calculate dCA1 = CA1 C2 /(1 + r) I1 dr dr 1 + r + (1 + r) r (6)
where dr is as in eq. (5) above except for the obvious changes:

F (I1 ) + [1 + r + (1 + r) ] I1 A 2 I1
dr =
I1 (C2 + C2 ) [1 + r + (1 + r) ] + 1+r r r
! dA2 .
The necessary and sucient condition for CA1 to rise is clearly, from eq. (6) above, C2 I1 [1 + r + (1 + r) ] > CA1 . 1+r r But there is no reason this has to hold (we can guarantee it fails by making and I1 /r very small while making Y1 as large as it has to be to generate a positive CA1 of sucient magnitude). Summary: If Home has a current account surplus on date 1 when A2 rises, its date 1 surplus falls (and Foreigns date 1 decit shrinks, or course). If A rises in the same circum2 stances, Homes date 1 surplus may rise or fall and Foreigns date 1 decit (correspondingly) may rise or fall. 7. (a) The intertemporal Euler equation is exp(C1 /) = (/R) exp(C2 /). Taking logs and solving yields C2 = C1 + log(/R). (b) By the intertemporal budget constraint, C2 = (1/R)(Y1 C1 ) + Y2 . Substituting this into the preceding Euler equation and solving yields C1 = W1 R log(/R) Y1 + RY2 R log(/R) = . 1+R 1+R 9
(c) Total dierentiation with respect to R yields dC1 Y2 log(/R) + W1 R log(/R) = dR 1+R (1 + R)2 C1 Y2 = + + [1 log(/R)] . 1+R 1+R 1+R (d) We have (C) = exp(C/) = . (C/) exp(C/) C
(e) Using the Euler equation in part a, one can substitute for C1 above to get dC1 Y2 C2 (C2 )C2 = + . dR 1+R 1+R What do these terms capture? The rst is the terms-of-trade eect, i.e., the wealth eect on C1 , Y2 /(1 + R), plus the Slutsky income-eect term (C1 /W1 )C2 = C2 /(1 + R). [See eq. (30) in Chapter 1.] To see that the nal term is the compensated price response of demand (i.e., the pure substitution eect), use the Euler equation to write lifetime utility as U1 = exp(C1 /) exp [C1 / log(/R)] = (1 + R) exp(C1 /) and solve for C1 as a function of R and U1 :
h C1 (R, U1 ) = log [(1 + R)] log(U1 ).
From this equation,
h (C2 )C2 C1 = = . R 1+R 1+R
8. (a) The Home government maximizes u(C1 , C2 ) subject to the national constraint C2 = Y2 + (1 + r)(Y1 C1 ) where r is the world interest rate. 10
Since private Foreigners carry out a similar maximization of u[C1 , Y2 + (1 + r)(Y1 C1 )], their date 1 consumption obeys the Euler equation u [C1 , Y2 + (1 + r)(Y1 C1 )] /C1
= (1 + r)u [C1 , Y2 + (1 + r)(Y1 C1 )] /C2 , which implicitly yields the function C1 (r). Thus, we may regard the Home government as maximizing (with respect to C1 )
u [C1 , Y2 + (1 + r)(Y1 C1 )] subject to the goods-market equilibrium condition

C1 + C1 (r) = Y1 + Y1 ,
which implies that C1 will inuence r. Indeed, by substitution into the Home welfare criterion of this last constraint, we may view the Home government as directly choosing the world interest rate r to maximize
u {Y1 + Y1 C1 (r), Y2 + (1 + r) [C1 (r) Y1 ]} .
The rst-order condition for this last (unconstrained) maximization with respect to r is: du u(C1 , C2 ) u(C1 , C2 ) u(C1 , C2 ) 0 0 C1 (r) + (1 + r)C1 (r) + [C1 (r)Y1 ] = dr C1 C2 C2 = 0. The trick now is to recognize that if the government is manipulating the world interest rate through a borrowing tax , the Euler equation for Homes private agents will be u(C1 , C2 ) u(C1 , C2 ) = (1 + r)(1 + ) , C1 C2 11
so that we may write the Home governments Euler equation (the displayed equation that precedes the Home private Euler equation) as du 0 = (1 + r)C1 (r)dr + [C1 (r) Y1 ] dr = 0 [u(C1 , C2 )/C2 ] (7)
after division by u(C1 , C2 )/C2 . Remembering that the country wishes to lower its rst-period borrowing cost (1 + r) [Y1 C1 (r)] > 0 by reducing the world real interest rate (dr < 0), we see that the second summand above, [C1 (r) Y1 ] dr > 0, is the direct gain in real income from a lower borrowing cost. The rst summand is a welfare loss due to the fact that rst-period borrowing from Foreigners is being reduced [Foreign consumption is rising, 0 C1 (r)dr > 0], despite the fact that the tax wedge makes the domestic marginal utility of rst-period consumption higher than its true, world opportunity cost. As usual, the optimal tax exactly balances the marginal domestic terms-of-trade gain against the marginal domestic welfare loss. The rest is merely a matter of manipulating denitions. Recall that IM2 C2 Y2 (there is no investment in this example), so that the Foreign budget constraint yields IM2 (r) = (1 + r) [Y1 C1 (r)] . Therefore,
0 (1 + r)IM2 (r)/IM2 (r) = 0 (1 + r)C1 (r) + [Y1 C1 (r)] Y1 C1 (r)
1 = 1+ , where is the optimal tax derived from solving the Home government rstorder condition, eq. (7). This last equation, however, implies that = as part a of this problem asserts. 1 , 1
(b) The condition > 1 can be understood as follows. If Foreign demand for second-period imports of consumption is inelastic ( 1), then a rise 12
in their price (a fall in the world interest rate 1 + r) elicits a proportionally smaller fall in demand. But since IM2 (r) = (1 + r) [Y1 C1 (r)], this means that [Y1 C1 (r)] alone actually must rise (stay constant for = 1) when 1 + r falls. According to the reasoning following eq. (7), however, Home welfare thus unambiguously rises when 1 and the borrowing tax is raised. This means that a maximizing Home government will never nd itself at a position where 1. (Think of this condition as analogous to the familiar result that a monopolist always operates on the elastic stretch of the demand curve it faces.)
13
Workbook2
Chapter 2 Solutions
1. (a) The current account identity can be written as Bs+1 = (1+r)Bs +T Bs . Now just plug in the assumed trade balance rule. (b) Using the answer to part a, for any > 0,
X s=t
1 1+r
st
rBs
st 1 = r[1 + (1 )r]st Bt 1+r s=t rBt = = (1 + r)Bt . 1 + (1 )r 1 1+r
(c) Under the rule above, debt grows without bound if < 1. But once the debt is as big as Y /r, the country can honor its foreign commitments only if debt stops growing and consumption is zero forever. Thus, the suggested rule must entail negative consumption levels at some point, which are inadmissible. To see directly why, consider the constant-output case, in which T Bs = Y Cs = rBs so that the payback rule implies Cs = Y + rBs . Notice that since Bs , Cs must at some point become negative. The rule therefore is consistent with intertemporal solvency only if we counterfactually allow for negative consumption levels: the price of high consumption today would be infeasibly high trade surpluses later on. In general, suppose output grows at the gross rate 1 + g, so that Ys = (1 + g)st Yt . Unless 1 + g
1
14
is at least as great as the gross growth rate of debt, which was shown to be 1 + r(1 ) in part a, the external debt-output ratio is unbounded. Thus the minimal payback fraction consistent with intertemporal solvency and positive consumption is = 1 (g/r) (which is positive if we assume that g < r). 2. (a) The expected utility Et Ut is a weighted average over dierent life spans, with weights equal to the survival probabilities: Et Ut = (1 ) [u(Ct )] + (1 ) [u(Ct ) + u(Ct+1 )] +
h i
+2 (1 ) u(Ct ) + u(Ct+1 ) + 2 u(Ct+2 ) + ....
(b) The result follows simply by expanding the expression in part a and grouping terms together. 3. Recall that with isoelastic utility, C 1 1 , > 0, u(C) = 1 log C, = 1.
(
1
Using the intertemporal Euler equation, we thus obtain, (1 + r) = 1 = Et 1 Ct+1 Ct

1/ ) .
(1)
Since consumption has a conditional lognormal distribution, the natural log of the gross consumption growth rate is conditionally normally distributed: Ct+1 Ct+1 Ct+1 log N Et log , Vart log Ct Ct Ct Thus
C 1
Et
t+1
Ct
Ct+1 1 = Et exp log Ct

1 Ct+1 1 Ct+1 = exp Et log + 2 Vart log Ct 2 Ct 15
. (2)
[Consult footnote 41 on p. 313 of the book. Equation (2) follows from computing the mean and variance of the random variable (1/) log(Ct+1 /Ct ), which is normally distributed when log(Ct+1 /Ct ) is.] Combining eqs. (1) and (2) above and taking natural logs of the result, we arrive at Et log or log Ct+1 log Ct =
Ct+1 Ct
1 Ct+1 Vart log 2 Ct
1 Vart {t+1 } + t+1 , 2
where t+1 log Ct+1 Et {log Ct+1 }. Since t+1 is a normal random variable that is uncorrelated with past information (because it is a pure forecast error), it is also statistically independent of that information on the assumption that the past information itself is generated by a jointly normal (i.e., Gaussian) stochastic process. In that case the conditional variance in the preceding equation actually is a time-invariant constant, so the natural log of consumption follows a random walk with a constant drift equal to 1 Var {t+1 } . 2 4. (a) Using eq. (32) in Chapter 2, we can write Ct+1 Ct = r(Bt+1 Bt ) +
X
The current account identity gives
1 r 1 + r s=t+1 1 + r
s(t+1)
Et+1 Ys
X s=t
1 1+r
st
Et Ys .
1 r X Bt+1 Bt = Yt + rBt Ct = Yt 1 + r s=t 1 + r
st
Et Ys ,
which can be substituted into the previous equation for consumption to give the result that the change in consumption equals the present value of changes in expected future output levels.
16
(b) If the process for output follows Yt+1 Yt = (Yt Yt1 ) + t+1 , then (Et+1 Et )Yt+1 = t+1 , (Et+1 Et )Yt+2 = (1 + )t+1 , (Et+1 Et )Yt+3 = (1 + + 2 )t+1 , and so on. Therefore, for s > t, (Et+1 Et )Ys = 1 st t+1 . 1
(c) Substituting the last expression into the equation for the change in consumption derived in part a, we get the following Ct+1 Ct 1 1 1 r t+1 + ... = + 1+r 1 1 1+r # 2 ... 1 (1 + r)(1 ) " !# r (1 + r) 1+r t+1 = 1+r (1 )r 1 1 + r 1+r t+1 . = 1+r
" ! !
(3)
As a result, provided that 0 < < 1, the desire to smooth consumption makes consumption innovations more variable than output innovations. (d) The current account identity for date t + 1 is CAt+1 = Bt+2 Bt+1 = Yt+1 + rBt+1 Ct+1 . Because Yt+1 Et Yt+1 = t+1 and, by eq. (3) from part c above, Ct+1 Et Ct+1 = Ct+1 Ct = 1+r t+1 , 1+r
the preceding current account identity gives a current account innovation of t+1 1+r t+1 = t+1 < 0. 1+r 1+r 17
Thus, a positive output innovation leads to a current account decit, as claimed at the end of section 2.3.3 in the book. 5. Work backward from the equation CAt+1 Zt+1 (1 + r)CAt = t+1 , where t+1 is uncorrelated with date t or earlier information. Taking expectations with respect to date t information yields Et CAt+1 Et Zt+1 (1 + r)CAt = 0. The previous equation can be rearranged to express CAt as CAt = 1 1 Et CAt+1 Et Zt+1 . 1+r 1+r
Through forward recursive substitution (and using the law of iterated conditional expectations ) we obtain CAt =
j
X
s=t+1
1 1+r
st
Et Zs
1 [because as j , 1+r Et CAt+j 0]. This is Campbells (1987) saving for a rainy day equation, eq. (43) in Chapter 2. The equation can alterna-
tively be derived using the lag and lead operator methodology described in supplement C to Chapter 2. Start again with CAt+1 Zt+1 (1 + r)CAt = t+1 and take expectations with respect to date t information to get Et CAt+1 Et Zt+1 (1 + r)CAt = 0. Using the lead operator we write this as L1 CAt L1 Zt (1 + r)CAt = 0, 18
or, dividing by 1 + r and rearranging, as
1 1 L1 CAt = L1 Zt . 1+r 1+r

Inversion of the lag polynomial on the left-hand side above gives CAt
1 1 1 = 1 Et Zt+1 L1 1+r 1+r st 1 X 1 = Et Zs+1 1 + r s=t 1 + r X 1 st = Et Zs . s=t+1 1 + r
To derive the converse, that the last equation implies CAt+1 Zt+1 (1 + r)CAt = t+1 , one can simply reverse the steps above. 6. Write the expression for the current account as follows CAt
X 1 r Et = Zt Et t = Zt 1 + r s=t 1 + r 1 r 1 1 = Zt 1 Et Zt , L 1+r 1+r
e Z
st
Zs
where the last equality is suggested in the hint. Then, multiplying both sides 1 by 1 1+r L1 , we have
1 1 1 L1 CAt = (Et Zt+1 Zt ) = L1 Et Zt 1 1+r 1+r 1+r

Therefore, CAt
st X 1 1 1 L Et = Zs 1+r s=t 1 + r X 1 st = Et Zs . s=t+1 1 + r
19
7. Equation (75) in appendix 2A of the book shows that as time T passes and as B/Y along its unstable increasing trajectory, Bt+T +1 /Yt+T +1 (1 + r) . Bt+T /Yt+T 1+g Because Ys+1 /Ys = 1 + g, however, it follows that Bt+T +1 /Bt+T (1 + r) as T , that is, net foreign assets grow asymptotically at the growth rate of consumption. But the gross rate of consumption growth, (1 + r) , must be strictly below 1 + r if, as we have assumed, an individual optimal plan exists (see the discussion on p. 117 of the book). Thus the asymptotic gross growth rate of foreign assets is below 1 + r and the transversality condition
T
lim (1 + r)T Bt+T +1 = 0
therefore is satised. 8. If permanent output uctuates more than current output, as is the case when output is a nonstationary random variable, then, as shown in exercise 4(d), a positive output innovation implies a decline in the current account: the intertemporal approach can therefore yield a countercyclical current account. Also, in the presence of investment which enters the current account with a negative sign, the current account can worsen following a positive output innovation if is suciently large. Refer to pp. 86-87 in the book for details. 9. (a) Dierentiating the rms objective function with respect to Is and Ks , we get, respectively, qs = 1 + Is and qs = 1 [As+1 F 0 (Ks+1 ) + qs+1 ] . 1+r 20
(b) Combining the preceding two rst-order conditions, we get the following dynamic system for q and K: Kt+1 Kt = qt+1 qt = rqt At+1 F qt 1 ,
0
qt 1 Kt + .
(c) The phase diagram looks qualitatively the same as gure 2.9 in the book, which is based on eqs. (66) and (67) in Chapter 2. The dynamics of q and K and the slopes of the corresponding schedules are, however, quantitatively dierent. For example, the slope of the q = 0 schedule now is dq dK q=0
q1 AFKK K + ! < 0, = q1 1 r AFKK K +
not just in a neighborhood of the steady state, but globally. (Compare with the slope given on p. 109 of the book.) As in the slightly more complex q model of Chapter 2, the steady state is given by q = 1, AFK (K, L) = r. The steady state is independent of the adjustment costs, which determine only the speed of transition to the steady state. (d) Figure 2.9 can be used for the exercise. An unanticipated permanent rise in A shifts the q = 0 schedule immediately and permanently to the right, raising the steady state capital stock to K 0 . The unique convergent saddlepath SS also shifts to the right, becoming S0 S0 . Because the initial capital stock is given as K < K 0 , q rises in the short run (to place the economy on its new saddle-path) and investment surges. Over time, however, q falls back to 1 and investment decreases as K K 0 . (e) The principle for analyzing anticipated shocks is the same as that applied in gure 2.11. The rm learns on date t that productivity A will rise permanently to A0 at a known future date T. Thus, the shifts described in 21
schedules described in part d occur only on date T . Nonetheless, the rm will adjust in anticipation so as to smooth its investment costs; and between dates t and T , q and K therefore will follow the original equations of motion (those involving A rather than A0 ). Thus, q jumps up initially and, until date T , continues rising as capital is accumulated. The rm reaches the new saddle-path S0 S0 precisely on date T , and thereafter q falls toward 1 and K rises to its new steady state. (An anticipated future fall in A would induce a path qualitatively similar to the path shown in gure 2.11.) (f) Marginal q does not equal average q because the installation cost function assumed in this exercise is not linear homogeneous in K and I.
22
Workbook2
Chapter 3 Solutions
1. (a) Because r = 0, an individuals desired consumption when young would be 1 1 cy = [y y + (1 + e) y y ] = (2 + e) y y 3 3 if unrestricted borrowing were possible. In general 1 c = min y , (2 + e) y y . 3
y y
Obviously the borrowing constraint will bind only when e > 1. Since = 1, co = cm = cy when the borrowing constraint of the young doesnt bind. If it does (that is, if e > 1), then co = cm = The saving of a young person is 1 1e y y . s = max 0, y (2 + e) y y = max 0, 3 3
y y
1 (1 + e) y y . 2

That of a middle-aged person is sm = (1 + e) y y

1
(1 + e) y y sy (1 + e) y y + sy = , 2 2
23
and that of an old person is

o y m
(1 + e) y y + sy . s = (s + s ) = 2 (b) Now 1 + g is the gross growth rate of a young-persons output. Aggregate saving out of total output is sy + sm + so t t t , y o yt + yt or Aggregate = saving rate

y y (1+e) yt2 + sy (1+e) yt1 sy 1e y t1 t2 max 0, yt + 3 2 2 y [(1 + g) + (1 + e)] yt1
"
"
(1) Let us rst assume that e 1, so that the young do not wish to borrow a positive amount. In that case, eq. (1) becomes 1e 1 + 2e y 2+e y y (1 + g)yt1 + yt1 y 3 3 3(1 + g) t1 Aggregate = y saving rate [(1 + g) + (1 + e)] yt1 = (1 e)(1 + g)2 + (1 + 2e)(1 + g) (2 + e) . 3 [(1 + g)2 + (1 + g)(1 + e)] (2)
Borrowing by the young becomes an issue only if e > 1. If the young cannot borrow sy = 0 and aggregate saving is computed accordingly, taking account of the fact sm and so are not the same as when the young can borrow freely. In the borrowing-constrained case, eq. (2) is replaced by Aggregate saving rate
y y (1 + e) yt1 (1 + e) yt2 sm + so 2 2 t = t = y o y yt + yt [(1 + g) + (1 + e)] yt1 (1 + e) (1 + e) 2 2(1 + g) = (1 + g) + (1 + e) g(1 + e) = . 2 + (1 + g)(1 + e)] 2 [(1 + g)
(3)
24
(c) Take the derivative with respect to e of the numerator of eq. (2). It is (1 + g)2 + 2(1 + g) 1 = g 2 < 0. Because, in addition, the denominator of (2) rises when e rises, steeper income growth between youth and middle age depresses saving. Intuitively, the young and old save less and the middle-aged save more, but with positive overall economic growth (g > 0), it is the eects on the young and old that dominate. When the young cant borrow and e > 1, the derivative is computed from eq. (3) and is proportional to 2(1 + g)g > 0. The preceding eect is reversed: the positive eect of a higher e on middle-age saving dominates. (d) Observe that 2 y 1 m sy = yt yt+1 , t 3 3 2 m 1 y sm = yt yt1 , t 3 3 1 y m so = yt2 + yt1 . t 3 Thus, in the rst case described in this section of the exercise, the aggregate saving rate is Aggregate = saving rate =
h
2 y (1+g)2 yt2 3
1 ym + 3
(1
y (1 + g)2 yt2 + y m
2 2 g 3
y + g yt2
2 m y y 1 (1+g)yt2 3 3 y + g)2 yt2 + y m
1 3
y yt2 + y m
Taking the derivative above with respect to g, we nd it is proportional to
4 4 1 y 1 + g y m + 1 + g + g 2 yt2 > 0. 3 3 3
Thus countries with higher output growth rates for young workers will have higher saving rates (all else equal). In the second case, in which growth is
25
concentrated on the middle-aged, Aggregate = saving rate =

h
2 y y 1 (1 3 3 m + g)2 yt1 +
m y y + (1 + g)yt1
2 (1 3
m m + g)yt1 1 y y 1 y y + yt1 3 3
m 1 g 2 yt1 3 . m y y + (1 + g)yt1
It is easy to see that the derivative with respect to g of the last expression is negative. Growth concentrated in middle-aged workers lowers the national saving rate if the young can borrow against future earnings. 2. (a) With log utility, cy = t wt , 1+ co = t+1 (1 + r)wt . 1+
The indirect utility function of a young agent is then U = (1 + ) log(w) + log(1 + r), apart from an irrelevant additive constant. Dierentiation yields dU = dr
1+ w
dw + . dr 1+r
For an innitesimal change in r, using the envelope theorem we have that dw/dr = k, so that dU 1+ = . k+ dr w 1+r
!
(4)
A representative young agent saves sy = wt wt /(1 + ) = wt /(1 + ) on t date t. Thus, starting from a situation where the world and autarky interest rates are equal, we have Kt+1 Nt sy t = , Nt+1 Nt+1 26
which can be rewritten in per capita terms as kt+1 = wt sy t = . (1 + n) (1 + )(1 + n)
Substituting for k in eq. (4) yields dU = + . dr 1+n 1+r Hence dU/dr > 0and the current young and all future generations benet from a rise in r on date tif and only if r < n, as the exercise assumes. [Because saving per worker is constant after the change and k = sy /(1 + n), we can express the lifetime income change for a new generation as kdr + 1+n kdr > 0.] The date t old benet because of a higher return on their 1+r previous saving. (b) Opening the economy to trade will bring the domestic autarky interest rate down to the world level. Suppose the world rate r is innitesimally below ra . Then the change makes all generations worse o if ra < n. The problem is that we are removing one distortion, trade barriers, while leaving dynamic ineciency uncorrected. Further, opening to trade exacerbates dynamic ineciency because r < ra < n. (The result is not true in general for non-innitesimal gaps between the autarky and world interest rates. Closed-economy intuition may be misleading in this case, because in an open economy, as r falls the economy does not have to save more to maintain an ever-higher capital-labor ratio. It borrows for that purpose instead, and in the dynamically inecient case, higher steady-state per capita foreign debt implies highernot lowersteady state per capita consumption.) 3. [There are two typos in the statement of this exercise. One line up from the bottom of p. 195, u(ct ) should be u(cs ). On p. 196 in part c, the term v v v (yt tv ) should be (ys s ).] (a) On any date t, all the members of the generation born on t still are alive, and contribute 1 to population. Of those born on t 1, 1 = remain on 27
date t. Only (1 ) = 2 of those born on date t 2 are still around on date t. And so on. Total population on any date therefore is 1 + + 2 + 3 + ... = 1/(1 ). (b) The insurance industry pays a gross return of (1 + r)/ to those who actually survive the period, but nothing to those who dont. Only a fraction of the existing population makes it from date t into date t + 1. Thus the industrys gross payout on the assets B (positive or negative) that it holds is 1+r B = (1 + r)B,
which exactly equals its earnings. Hence prots are zero. (c) Given the eective interest rate individuals face, the asset-accumulation identity is 1 + r p,v v bp,v = b + yt tv cv , (5) t+1 t t or, in terms of the lag operator, 1 + r p,v v L1 ) bt = cv (yt tv ). (1 t 1+r Invert the left-hand side lag polynomial above and impose the condition T limT 1+r bp,v +1 = 0. The result is the intertemporal budget constraint t+T
1 + r p,v X bt = s=t 1 + r
st
v v [cv (ys s )] . s
(d) We know that for log consumption the individuals optimum plan is cv = (1 ) t
"
1 + r p,v X bt + s=t 1 + r
st
v v (ys s )
(6)
(because the eective subjective discount factor is ). The aggregated version of eq. (5) is slightly intricate to derive, but we use the parenthetical hint 28
at the end of the question for this part. Recall that bp,v is the end-of-(t 1) t assets of someone from vintage v. Thus for the economy as a whole,
p,t1 p,t2 p + bt + 2 bp,t3 + ..., Bt = 1 bt t
p,t1 p,t2 p Bt+1 = 1 bp,t + bt+1 + 2 bt+1 + ..., t+1
while, in contrast,
t t1 t2 Yt = 1 yt + yt + 2 yt + ....,
etc. Being careful about time subscripts, we therefore may aggregate eq. (6) as Ct = =
h p,t1 1+r 1 bp,t + bt + 2 t st (1 ) P
p,t2 bt + ...
= (1 ) (1 +
h p,t1 p,t2 1+r 1 bt + bt st (1 ) P
s=t
1+r
(Ys Ts )
"
s=t
p r) Bt
1+r X s=t
1+r
(Ys Ts )
st
i + ... #
(Ys Ts ) ,
(7)
where the fact that bp,t = 0 (the newly born have no nancial wealth at the t start of the period they are born) has been used. Similarly, we aggregate eq. (5) as
p Bt+1 =
i (1 + r) h p,t1 p,t2 1 bp,t + bt + 2 bt + ... + Yt Tt Ct t i (1 + r) h p,t1 p,t2 1 bt = + bt + ... + Yt Tt Ct p = (1 + r)Bt + Yt Tt Ct .
p,t1 p,t2 1 bp,t + bt+1 + 2 bt+1 + ... t+1
The intuition for this relation (which is the usual one) is that the economy as a whole is earning the interest rate r on its net foreign assets. 29
(e) If Y and T are constants, eq. (7) can be substituted into the last aggregate equation to yield
p Bt+1
= (1 + = (1 +
p r)Bt
+ Y T (1 ) (1 +
" #
"
p r) Bt
p r)Bt
(1 + r) 1 + (Y T ). 1+r
(1 + r) (Y T ) + 1+r
The dynamics are qualitatively the same those shown in gure 3.9, assuming that the system is stable [meaning that 1 > (1 + r)]. The slope of the atter of the two diagonal lines is now (1 + r), however, and the lines vertical intercept is (1+r)1 (Y T ). 1+r (f) We can solve for steady state private assets using the equation in part e. p p We get, setting Bt+1 = Bt = B p , Bp = (1 + r) 1 (Y T ), (1 + r ) [1 (1 + r)]
!
and, for steady state aggregate consumption, C=
1 1+r
1 (1 + r)(Y T ). 1 (1 + r)
By substituting T = rD above, we can see the eects of a steady state public debt of D. Clearly a higher D depresses steady-state consumption, from the last equation, assuming the dynamic stability condition (which is needed for positive steady state consumption). Since output is exogenous, this must mean that steady-state total net foreign assetsthe sum of government and private net foreign assets are lower. That is, private net foreign assets rise by less than government debt. Note that in the Ricardian case ( = 1), we have, in contrast, Y B p = + D, r so that a rise in D raises steady-state net private foreign assets one-for-one.
30
4. In the case of a dynamically inecient world economy with r < n (see section 3.6.4), the introduction of a public debt nanced entirely by taxes on the young can raise the welfare of all generations, and in both countries too! If the two countries initially nance their capital stocks entirely out of their own savings, with no net asset trade, then the introduction of a small public debt in Home reduces the worlds dynamic ineciency without any international redistribution eects. As in exercise 2, part a, above, everyone in the world benets from the world interest rate rise, and Homes young pay lower taxes to their government as explained in section 3.6.4. Net international private lending changes the analysis, however, by introducing redistribution eects due to intertemporal terms of trade changes. When Home is initially a net creditor of Foreign, for example, the interest rate rise creates a further termsof-trade benet for Home, but at Foreigns expense. So Foreigns current and future generations may be net losers despite the gains in Home. See section 3.5 on such international redistribution eects. 5. Intuitively, we know that (1 + r)ndt /r (where n < r) is the contribution of the existing public debt to the net wealth of vintages alive on date t. Any further debt issues ds+1 ds will raise net wealth on date s by the amount (1 + r)n(ds+1 ds )/r. This explains the consumption function you are asked to derive. To derive the expression formally, rewrite the government nance constraint as t = (1 + r)dt (1 + n)dt+1 + gt 1+r 1 1+r (r n) dt dt+1 n(dt+1 dt ) + gt . = r 1+r r (Dont be depressed if this last equality isnt immediately obvious. But do be sure to combine terms to check its validity.) Let L1 denote the lead operator, as usual (see supplement C to Chapter 2), and write the last equality as t =
h i 1+r 1+r (r n) 1 (1 + r)1 L1 dt n(dt+1 dt ) + gt . r r
(8)
31
Observe next that the present discounted value of per capita current and future taxes can be expressed as
X s=t
1 1+r
st
s = 1 (1 + r)1 L1
i1
t .
Substituting for taxes using formula (8), we nd that:

X s=t
1 1+r
st
X 1 1+r (rn)dt + s = r s=t 1 + r
st
1+r n (ds+1 ds ) . gs r
Substitution of the above equation for the present value of aggregate per capita taxes in eq. (66) of Chapter 3 yields the desired result. One also has to make use of the identity b = bp d (i.e., the economys total net foreign assets are the sum of public net assets and private net assets). 6. The government is not indierent as to the path of distorting taxes, because dierent tax paths of equal present value may inict dierent total distortion costs on the economy. To see this, solve the governments planning problem. The government picks the paths of private consumption and taxes to maximize X 1 st Ut = u(Cs ) s=t 1 + r subject to the consumers rule for intertemporal consumption smoothing, u0 (Cs ) = u0 (Cs+1 ), and subject to the private-sector budget constraint,
X s=t
1 1+r
st
aT 2 Ys s Ts Cs = 0, 2
and the government constraint

X s=t
1 1+r
st
(Ts Gs ) = 0.
32
(For simplicity a zero date t public debt is assumed.) The private Euler equation implies that desired consumption is at at some level C. Thus, we can write the Lagrangian for the governments problem as
i X 1+r h 1 Lt = u(C) C r s=t 1 + r
st "
aT 2 Ys s Ts + (Ts Gs ) 2
where is the Lagrange multiplier on the private consumption constraint and that on the government budget constraint. The rst order conditions, found by dierentiating with respect to C and Ts (s t) are: u0 (C) = , u0 (C) = u0 (C)aTs . The second of these conditions states that at an optimum, the shadow value of government resources, , exceeds their private value, u0 (C), by an amount equal to the marginal deadweight loss inicted by the tax (measured in util ity). Because both and u0 (C) are constant over time, however, taxes are constant over time as well. In other words, the government nds it optimal to smooth taxes, just as consumers smooth consumption. The constant level of tax is 0 = u (C) . T au0 (C) One solves for using the government budget constraint,
X s=t
1 1+r
st "
u0 (C) Gs = 0, au0 (C)

st
to nd = u (C) +
0
[Interpretation: the government shadow value of revenue exceeds u0 (C) by a weighted average of current and future marginal consumption costs due to the exogenous stream of public expenditures.] Since there is an optimal level of taxes, there is also an optimal public decit on each date, in contrast to Ricardian equivalence models. When G is unusually high, the government 33
X 1 r u0 (C) 1+r s=t 1 + r
aGs .
will run a decit rather than raising taxes, and when G is unusually low it will run a surplus.
34
Workbook2
Chapter 4 Solutions
1. We follow closely the steps outlined in section 4.3.2. Equations (12) and (17) in the book remain unaltered. So does steady-state budget constraint b (11), so that Z = l w, as before. In the traded goods sector the zero prot condition can be written as f
kt Et
kt kt w ,1 = r + Et Et Et
(1)
where k K/L. Log-dierentiation of eq. (1) yields, f

0
kt Et
kt w kt kt Et = r kt Et + w Et . Et Et Et
0
(2)
The rst order condition for capital in the traded goods is f (kt /Et ) = r. Substitution of this equality into eq. (2) above reduces it to w = Et .
b Since Z = l w, the last equality implies
(3)
Log dierentiating the zero-prot condition for nontradables, pAn g(kn ) = rkn + w, with An and r constant, we obtain p = ln w [recall the displayed
1
b Z = l Et .
(4)
35
equation following eq. (8) in Chapter 4]. Since ln = 1 in section 4.3.2, eq. (3) above implies p = (1 )Et Substituting eqs. (4) and (5) above into equation (17) of Chapter 4, we express the change in steady state nontradables consumption as
b C n = {l (1 ) [ + (1 )]} Et .
b Similarly we can solve for Ln as,
b The sign of Ln is still ambiguous, as discussed on p. 224 in the book. Indeed, eq. (6) here looks precisely like eq. (18) of Chapter 4, with the only dier ence that Et appears in place of At /lt . Thus, the assumption of Harrodneutral instead of Hicks-neutral technological change makes no dierence for the likelihood that faster productivity growth causes a labor exodus from the tradables sector.
b Ln = {l (1 ) [ + (1 )] } Et .
(6)
2. This is an intricate problem, so it is useful to start with an exceedingly simple case and gradually generalize it. First, suppose that consumption preferences are of the Leontief type (with xed consumption proportions of the two goods) and that technologies also are Leontief. Since nontradables are relatively labor intensive, a rise in r lowers their relative price in terms of tradables, p, and lowers the tradables wage, w. In the special case were now considering, the sole eect of that change is to atten the GNP line (by lowering p) and to shift its vertical intercept: upward if w0 (r)L + Q > 0, downward in the opposite case (the rst case occurring, despite the inequality w0 (r) < 0, if nancial wealth is suciently high). The income expansion path stays put due to the assumed Leontief preferences. How can we be certain the GDP line doesnt shift? Equations (7) and (10) of Chapter 4 allow us to write the GDP line as: ( ) At f [kt (r)] Yt = Yn + At f [kt (r)]L. (7) An g[kn (r)] 36
0 0 Since kt (r) = kn (r) = 0 with a Leontief technology, the interest-rate change doesnt shift the GDP line. The diagram therefore predicts that in the case of an upward shift in the GNP line (a positive national income eect), consumption of tradables and nontradables expands as the economy reduces its capital stock and acquires the corresponding amount of foreign assets. In the case of a downward shift in the GNP lines vertical intercept, the eects are opposite. (Henceforth assume an upward shift to avoid a proliferation of cases.) Now, as a second, slightly more general case, allow substitution in consumption between tradables and nontradables. In that case a rise in r, by lowering p, makes the income expansion path rotate clockwise from the origin. This movement, other things the same, induces more consumption of nontradables, less of tradables, and a bigger shift out of domestic capital into foreign assets. Finally, what if we now relax the assumption of Leontief technologies? It is easy to see from eq. (7) above that while the change in the GDP lines slope is ambiguous, its vertical and horizontal 0 0 intercepts both decrease (because kt (r) < 0 and kn (r) < 0 now). Thus, the GDP line shifts inward. This last shift has no eect on the economys consumption point, which remains at the intersection of the GNP line and income expansion path. However, the shift implies a reduction in tradables output, and hence a further fall in domestic capital and rise in net foreign assets B. Because production is now becoming less capital-intensive in both sectors but output (equals consumption) of nontradables is the same, labor must ow out of tradables, reinforcing the economys disinvestment of capital. Accordingly, more of the economys tradables consumption is nanced out of foreign investment income rather than domestic production. The three-step solution oered here corresponds to three distinct eects of the rise in r: a budget-constraint eect, a consumption switching eect, and a capital-intensity eect.
3. (There is a typo in the statement of this question. The second to last sentence should ask for the change in Bt+1 /Pt , not the change in Bt /Pt .) Divide 37
both sides of eq. (30) in the book by Pt . Next, manipulating the resulting equality as is done on p. 230, section 4.4.1.4, we obtain an alternative version of eq. (27):
X s=t c Rt,s Cs = c (1 + rt )Bt X c Yt,s + ps Yn,s Is Gs + Rt,s . Pt1 Ps s=t
(8)
The Euler equation for C, when period utility is isoelastic with intertemporal substitution elasticity , is eq. (33) in the chapter. Through recursive substitution, optimal Cs follows:
c Cs = (Rt,s ) (st) Ct .
Now substitute the preceding equation into (8) to obtain the equation for optimal real consumption: Ct =
c (1+rt )Bt Pt1
The current account identity is
(st) Rc t,s s=t
s=t
n,s c Rt,s Yt,s +ps YPs Is Gs
CAt = Bt+1 Bt = rBt + Yt,t + pt Yn,t Pt Ct It Gt = rBt + Zt Pt Ct , where Zt Yt,t + pt Yn,t It Gt . Divide the current account identity by Pt (to put it in terms of real consumption). The result is Bt+1 (1 + r)Bt Zt (1 + r)Pt1 Bt Zt = + Ct = + Ct , Pt Pt Pt Pt Pt1 Pt which can be rewritten as Bt+1 Bt rc Bt Zt = t + Ct . Pt Pt1 Pt1 Pt Substitute into this the equation for consumption derived above. The result is c P (1+rt )Bt c s c + Rt,s Zs Bt+1 Bt rt Bt Zt s=t Pt1 P = + P 1 . Pt Pt1 Pt1 Pt (st) Rc
s=t t,s
38
As in section 2.2.2 of the book, let us dene t

P c 1 (st) s=t Rt,s . P c
s=t
Rt,s
Then we can write the last equation as
Bt rc Bt Zt Bt+1 = t + Pt Pt1 Pt1 Pt rc Bt Zt = t + Pt1 Pt t 1 + t

P
c (1+rt )Bt Pt1
+ +
c (1+rt )Bt Pt1
Ps=t c t Rt,s
P
s=t
c s Rt,s Zs P
c ! (1+rt )Bt
Pt1
Dene the permanent level of a variable as

s=t R Xs Xt = P t,sc . s=t Rt,s c
c Zs s=t Rt,s Ps P c s=t Rt,s
c Zs s=t Rt,s Ps P c s=t Rt,s
Then the last equation can be expressed as the analog of eq. (26) in section 2.2.2,
g Bt Bt Zt Bt+1 t 1 Z c c = (rt rt ) + + Pt Pt1 Pt1 Pt P t t !" g rt Bt c Z + Pt1 P #
t
where we have used the following relation (from footnote 13 on p. 78 of the book): c 1 + rt = rt . c P c s=t Rt,s 4. (a) With a constant world interest rate r equal to (1 )/ and a constant net supply of nontradables, consumption of tradables is constant too, and equal to the level specied in eq. (10) of Chapter 2, provided we reinterpret Y, G, and I in that equation as output, government consumption, and investment of tradable goods. (The book already assumes that only 39
tradable goods can be transformed into capital.) To see this, consider eq. (35) in Chapter 4, and note that the result claimed follows [with r = (1 )/] provided the price index P is constant over time. By eq. (20) in Chapter 4, however, P depends only on p (the price of nontradables in terms of tradables); furthermore, the individuals static rst-order condition for optimal consumption is Cn (Ct , Cn ) = p. Ct (Ct , Cn )
Equation (34) (in the present case) can be written as Ps+1 Ct,s+1 = Ps Ct,s . Since Cn is constant over time (at Yn Gn ), eq. (34) therefore can be written as (Ct,s+1 , Yn Gn ) = (Ct,s , Yn Gn )
for the appropriately dened function (Ct , Cn ). But this last equation implies that Ct is constant over time, as must also be p and P . In this case, therefore, eq. (35) implies that optimal consumption of tradables is the annuity value of the net endowment of tradables after investment. (b) The previous result no longer holds when Yn Gn is subject to anticipated changes, except in the special case = . Lets call that case the benchmark case; all changes in consumption and the current account in the rest of this answer are relative to the economys path in the benchmark case. The argument in the answer to part a shows that consumption of tradables is constant until date t+T , when Yn Gn rises permanently to Yn0 G0n . From date t+T on consumption of tradables is constant again, after possibly jumping between dates t + T 1 and t + T . The current account is zero from date t+T on. There are two cases to consider in solving for the pre-t+T path (see p. 234 of the text for a discussion of the tension between intertemporal and intratemporal substitution eects): > . The Euler equation for Ct between dates t + T 1 and t + T is Ct,t+T =
!
Pt+T 1 Pt+T 40
Ct,t+T 1 ;
(9)
< . This case is a mirror image of the last one, with a current account decit prior to date t + T and Ct falling between dates t + T 1 and t + T as a result of the fall in P . [Once again, eq. (33) in the chapter precludes a rise in P between dates t + T 1 and t + T .]
see eq. (34) in Chapter 4. Since Yn Gn rises on date t + T , it is reasonable to make the tentative hypothesis that Pt+T is below Pt+T 1 , which implies via (9) above that Ct,t+T > Ct,t+T 1 . In that case Ct is constant and below its benchmark level prior to t + T , as the economy runs a current account surplus. The foreign assets accumulated in this way permit a higher level of tradables consumption from t + T on. Before concluding that we have solved this case we must ask whether P can ever rise on date t + T . This is impossible: a rise in P would require a (very large) upward jump in Ct , which would imply a violation of Euler eq. (33) in the chapter. (That equation, the Euler equation for total real consumption, implies that both Ct and Cn can rise between dates t + T 1 and t + T only if P falls.)
5. Departing from the notation in appendix 4A, let C denote the real consumption index depending on consumption of tradables and leisure, C = (Ct , L L). With leisure interpreted as consumption of nontradables, budget constraint (23) in the chapter can then be written as
X s=t
1 1+r
st h
Ct,s + ws (L Ls ) = (1 + r)Qt +
X s=t
1 1+r
st
(ws L Gt,s )
(recall that all government consumption is of tradables now). Constraint (24) therefore takes the form
X s=t
1 1+r
st
Ps Cs = (1 + r)Qt +
X s=t
1 1+r
st
(ws L Gt,s ).
Combining this equation with Euler eq. (33) from the chapter yields a real consumption function along the lines of (29): Ct = (1 + r)Qt + Pt
P
s=t
(1 + r)st 41
P 1 st (w
s=t 1+r
Pt Ps
i1
sL
Gt,s ) (st)
Applying eq. (22) in the chapter, we arrive at eq. (67). 6. (a) Think of the individual as maximizing the prot from undertaking education, which equals earnings less the unskilled labor income foregone. R The latter is w/(r + ), the former, T e(r+)t AT hdt, equal to the discounted value of earnings from human capital starting on the rst date after graduation, T . The rst-order condition is
Z d Z (r+)t e AT hdt = e(r+)T AT h + e(r+)t AT 1 hdt dT T T AT 1 h (r+)T = e(r+)T AT h + e = 0, r+
from which T = /(r + ) follows. Less sharply decreasing returns to education ( closer to 1) lengthens the optimal time in school, but a higher discount rate shortens optimal schooling. The reward h doesnt aect T here because it multiplies the term in the prot function that depends on T . (b) The rst-order condition is
Z d Z (r+)t e AT hdt = e(r+)T AT h + e(r+)t AT 1 hdt dT T T AT 1 h (r+)T = e(r+)T AT h + = 0, e r+
from which T = /(r + ) follows. Less sharply decreasing returns to education ( closer to 1) lengthens the optimal time in school, but a higher discount rate shortens optimal schooling. The reward h doesnt aect T here because it multiplies the term in the prot function that depends on T . (c) Using the solution for T , lifetime earnings are
Z
T
e(r+)t A(T ) hdt =
1 = A r+ r+
e(r+)t A
/(r+)
r+
hdt
he .
Equating this to w/(r + ) yields the answer that was claimed. 42
(d) Taking derivatives, dw = w + w + w log >0 d r+ provided, as assumed, > r+. (Hint: [/(r+)] = exp{ log[/(r+)]}.) The other derivatives are obvious, and so is the intuition. (e) If wages w are higher for the reasons in d, then since r is given, the relative price of nontradables p is higher. Since tradables sell at the same price worldwide, the real exchange rate is higher. This corresponds to the Harrod-Balassa-Samuelson result, as measured labor productivity in tradables is higher. 7. The representative individual in each country maximizes Ut = so dUt =
X s=t
st log Cs ,
X st C
s. Cs s=t As discussed in the chapter, the economy reaches its new steady state after one period. The change in lifetime utility relative to the baseline steady state path can therefore be written as b dUt = Ct + C. 1 s=t
X st dCs
Equations (55) and (59) in the chapter now allow us to substitute for r and b C: dUt 1 v = 2 1 = 1 1

4
Using eq. (51) in Chapter 4, we can write this sum as 1 b r C. dUt = (1 ) + 1

!
1 0 1 A
r v 1+r 2 1 A0 1 2 2
v . 2
43
[The last equality follows from r = (1 )/.] Because, A0 ( 1 ) < 0, however, 2 the preceding equation shows that dUt > 0. The temporary increase in Foreign productivity raises Homes lifetime utility. 8. (a) If Z is expenditure in imports, then demands are X = Z/p, M = (1 )Z. The price index P therefore satises (P/p) [(1 )P ]1 = 1, which implies the answer. (b) The proposed identity states that the current account (in real consumption units) equals output less consumption, both measured in like units. The intertemporal constraint is (1 + r)Bt +
X s=t
1 1+r
st
X ps Y 1 = Ps s=t 1 + r
st
ps Xs + Ms . Ps
(c) In terms of C the problem is to maximize Ut = subject to

X s=t X
1 1 st Cs 1 1 s=t
1 1+r
st
Cs = (1 + r)Bt +
X s=t
1 1+r
st
ps Y . Ps
The rst-order Euler condition therefore is the usual Cs+1 = (1 + r) Cs . In the notation of Chapter 2, we nd that
X r+ 1 Ct = (1 + r)Bt + 1+r s=t 1 + r
"
st
ps Y . Ps
Of course, optimal X = Z/p = P C/p, optimal M = (1 )P C, and, under the assumption made in the exercise, = 0.
44
(d) A temporary fall in p has eects on C qualitatively identical here to those of a temporary fall in Y , since p/P = (1 )1 p1 falls when p falls. The current account eect can be seen from Bt+1 Bt = rBt + pt Y /Pt Ct . Consumption C is constant after the initial fall in the terms of trade, and the country runs a current account decit while its terms of trade are temporarily depressed. (e) In this case the consumer maximizes Cs st Ut = 1 s=t subject to
X s=t X
1 1
1 1+r
st
Ps Cs = (1 + r)Bt +
X s=t
1 1+r
st
ps Y.
The rst-order Euler condition for C is now Cs+1 (1 + r)Ps = Ps+1

" #
Cs ,
which is formally identical to eq. (33) in Chapter 4. The dierence here, in comparison to the case analyzed in parts c and d, is that the expected future terms of trade change (occurring on date T , say) aects the path of the consumption-based real interest rate. We thus have a situation reminiscent of that in exercise 4 above. Since r = (1 )/, the Euler equation is Cs+1 = (Ps /Ps+1 ) Cs , so that P rises and C falls discretely the day the terms of trade switch back to their initial level. The current account (measured in imports) is given by Bt+1 Bt = rBt + pt Y Pt Ct . 45 (10)
To understand current-account behavior, notice that the Euler equation can be written as in terms of expenditure as: Ps+1 Cs+1 =
Ps Ps+1
!1
Ps Cs .
When 1, P C falls or remains unchanged (if = 0) when the terms of trade make their expected return up to their initial level on date T (that is, PT CT PT 1 CT 1 ). Since pY rises at the same time and the current account balance is zero starting on date T (because the expected future terms-of-trade path is again constant), we can infer from eq. (10) that the current account must have been in decit during the interval before T when the terms of trade were expected to improve. When < 1, however, we cannot make this argument, because P C rises on date T , and if P C rises then by more than pY , we would have to conclude that the current account was moving from a surplus to zero. Can this ever occur? It cannot, as we now show, provided it is assumed that the temporary fall in p harms the home country. (This need not be so, as we will discuss in closing.) Suppose that the economy runs a surplus between date 0, the date the terms of trade fall to p0 , and date T , when they return to p. Since foreign assets have risen between 0 and T , the economys residents are better o from T on than they were before date 0. The Euler equation tells, us, however, that CT = (PT 1 /PT ) CT 1 ; and, since PT > PT 1 , we see that CT 1 > CT , where CT is in turn higher than its level prior to date 0 as a result of the external surplus between dates 0 and T . So we get the following picture of what must happen if the economys response is a current account surplus: C rises when p rst falls and remains constant until date T , when C falls, but to a level permanently higher than before the initial fall in p. The implication is that the economy is better o as a result of the fall in p! If it is worse o, it must run a current account decit between dates 0 and T , as in the > 1 case. But how can a fall in the relative price of an economys export good ever make it better o? This is a logical possibility, it turns out, albeit a far-fetched 46
one. If the economy initially has a positive stock of net foreign assets B, then it eectively has a positive endowment rB of the import good. In that circumstance, a very large fall in p (the relative price of exports in terms of imports) can reduce the economys consumption of the initial import good and raise that of the initial export good enough that the economy becomes a net exporter of its previously imported good. In this implausible case of a trade pattern reversal, the fall in p can make the economy better o and result in a temporary surplus in the current account. Otherwise the more intuitive prediction of a decit is correct.
47
Workbook2
Chapter 5 Solutions
1. Since consumption on date 2 is the sum of endowment and payments of contingent assets for the realized state, we have B2 (s) = C2 (s) Y2 (s), s = 1, 2. For s = 1, premultiply by p(1)/(1 + r) and use eq. (16) in Chapter 5 to obtain the following: (1) p(1)Y2 (1) + p(2)Y2 (2) p(1) p(1) B2 (1) = Y2 (1) Y1 + (1 + r) 1+ 1+r 1+r " # p(1)Y2 (1) + p(2)Y2 (2) 1 = (1) Y1 (1) 1+ 1+ 1+r " # p(1) p(1)Y2 (1) + p(2)Y2 (2) Y2 (1) + (1) 1+r 1+r " # p(1)Y2 (1) + p(2)Y2 (2) 1 = (1) Y1 (1) 1+ 1+ 1+r p(1)Y2 (1) p(2)Y2 (2) (2) + (1) 1+r 1+r " # p(2)(2)Y2 (1) (1)/Y2 (1) p(1) = (1)CA1 + . 1+r (2)/Y2 (2) p(2)
1
"
48
Here, the last equality comes from eq. (17), Chapter 5. Using eq. (80) from the chapter, we see that the autarky price of the Arrow-Debreu security for state 1 relative to that of the state 2 security is p(1)a (1)/Y2 (1) = . p(2)a (2)/Y2 (2) Substitution of the preceding into the expression for p(1)B2 (1)/(1 + r), gives the required result. The result for B2 (2) follows from the identity CA1 = p(1) p(2) B2 (1) + B2 (2). 1+r 1+r
The statement of the exercise provides the intuition. 2. The necessary rst-order conditions are p(s) 0 0 u (C1 ) = (s)u [C2 (s)], s = 1, 2. 1+r For our utility function, u (C1 ) = 1/C1 and u [C2 (s)] = 1, so that the above conditions imply p(s) = (s)C1 , s = 1, 2. (1) 1+r (A similar relation holds for C1 , the initial consumption of Foreign residents.) If we divide eq. (1) for s = 1 by its analog for s = 2, we see that (1) p(1) = , (2) p(2) implying that equilibrium prices must be actuarially fair. Since p(1) +p(2) = 1, it follows that the Arrow-Debreu prices equal the respective probabilities of the state occurring: p(s) = (s), s = 1, 2. Assuming that Home and Foreign share the same discount factor , we may add eq. (1) for Home and for Foreign to obtain
C1 + C1 = Y1w =
0 0
2p(1) . (1 + r)(1)
49
Because p(1) = (1), we obtain an expression for the world interest rate 1+r = 2 . Y1w
Using Euler eq. (1) again, but substituting in this expression for 1 + r, we see that equilibrium date 1 consumptions are:
C1 = C1 =
Y1w . 2
The equilibrium intertemporal budget constraint of a Home resident is C1 + (1) (2) (1) (2) C2 (1) + C2 (2) = Y1 + Y2 (1) + Y2 (2), 1+r 1+r 1+r 1+r
so that date 2 consumptions must obey (1) (1) (2) Yw (2) C2 (1) + C2 (2) = Y1 1 + Y2 (1) + Y2 (2); 1+r 1+r 2 1+r 1+r (2)
there is a similar equation for Foreign. Since utility is linear in date 2 consumption with weights (1) and (2), a Home resident is indierent between any pair [C2 (1), C2 (2)] satisfying eq. (2). On date 2, however, goods-market equilibrium requires that
C2 (s) + C2 (s) = Y2w (s), s = 1, 2.
Thus, we have four equationseq. (2) and its Foreign analog, plus the state 1 and 2 equilibrium conditionsto determine the four unknowns [C2 (s), C2 (s)], s = 1, 2. If you play with these four equations, however, you will realize that the date 2 consumption allocation actually is indeterminate. To see this intuitively, imagine any equilibrium allocation of date consumption across states. Suppose now that Home were to reduce its state 1 consumption by 1/(1) units and raise its state 2 consumption by 1/(2) units. Constraint (2) would still be satised and Home residents would feel no worse o. If Foreign were simultaneously to raise its state 1 consumption by 1/(1) units 50
while lowering its state 2 consumption by 1/(2) unitsan action which is feasible for Foreign and does not lower its welfaredate 2 contingent commodity markets would still clear. Thus, the date 2 consumption allocation is not fully determined in the model. This indeterminacy result is, of course, a consequence of risk-neutrality with regard to date 2. People care about the expected values of second-period payos, but not about their distribution across states. This indierence explains why the national allocations of second-period consumption across the states of nature is not tied down. 3. (a) Ignoring nonnegativity, write the unconstrained maximization as a0 [(1 + r)B1 B2 + Y1 ]2 2 1 a0 2 E1 [(1 + r)B2 + Y2 (s)] [(1 + r)B2 + Y2 (s)] . + 1+r 2 max [(1 + r)B1 B2 + Y1 ]
B2
The rst-order condition for B2 is: C1 = E1 {C2 (s)} . The S + 1 budget constraints in the problem imply that E1
( ) ( )
C2 (s) C1 + 1+r
= E1
Y2 (s) (1 + r)B1 + Y1 + . 1+r

)
Thus by substitution of the rst-order condition,
1 Y2 (s) 1+ C1 = E1 (1 + r)B1 + Y1 + , 1+r 1+r 1+r Y2 (s) C1 = E1 (1 + r)B1 + Y1 + . 2+r 1+r

( )
that is,
(3)
For the -horizon case, we just get the usual permanent-income formula, essentially eq. (32) of Chapter 2 (suitably adapted). (b) The consumption formula of part a will be generally valid if the nonnegativity constraint on consumption never binds, that is, if, even when output 51
hits its minimal date 2 value (in state s = 1), C2 0. This last inequality will hold if and only if (1 + r)B2 + Y2 (1) 0 for B2 = (1 + r)B1 + Y1 C1 , where C1 is given by eq. (3) above. That is, we must have 1+r E1 Y2 (1 + r)B1 + Y1 + (1 + r) (1 + r)B1 + Y1 2+r 1+r which is equivalent to (1 + r)B1 + Y1 + 2+r Y2 (1) E1 Y2 . 1+r

+ Y2 (1) 0
If this inequality does not hold, then the nonnegativity constraint on C2 binds in at least one state of nature on date 2, so we cannot ignore the associated Kuhn-Tucker multiplier (see supplement A to Chapter 2). In that case, the Kuhn-Tucker theorem predicts that date 1 consumption must make C2 (1) = 0 (in state 1 of date 2 when output is minimal). Since C2 (1) = (1 + r) [(1 + r)B1 + Y1 C1 ] + Y2 (1) = 0 therefore holds, we see that C1 = (1 + r)B1 + Y1 + Y2 (1)/(1 + r). (c) The state-by-state Euler equations are p(s) (s) (1 a0 C1 ) = [1 a0 C2 (s)] , 1+r 1+r which reduce to C1 = C2 (s), s, because weve assumed p(s) = (s). Thus consumption is constant across states and dates, equal to C, given by 1+r C= 2+r
"
p(s)Y2 (s) 1+r Y1 + = 1+r 2+r s=1 52
S X
"
(s)Y2 (s) Y1 + . 1+r s=1
S X
The critical dierence between the equation above and eq. (3) is that the preceding equation holds ex post as well as ex ante, i.e., it holds in every state on date 2 as well as on date 1. Equation (3) above, in contrast, implies that date 2 consumption varies one-for-one with the output realization (date 2 consumption is not insured in the bonds-only asset regime). Thus the possibility of negative consumption is an issue in the bonds-only case, though not under complete markets. 4. Output-market equilibrium on date 1 requires that
N X n C1 = N N X 1 X n (Y1 + V1n ) = Y1n , 1 + n=1 n=1
n=1
or
n=1
N X
V1n =
n=1
N X
Y1n = Y1w .
(4)
(It is straightforward to check that if the preceding condition holds, the output market also clears on date 2, in every state s.) Next we have to nd equilibrium asset prices under condition (4), and check that they are indeed consistent with (4). Under (4) and the conjectured solutions for consumption, an agent in any country n has a marginal rate of substitution between date 1 consumption and date 2, state s consumption, of
N n u0 [C2 (s)] V1m C n Y w = n 1 = PN m=1 m = w1 . n u0 (C1 ) C2 (s) Y2 (s) m=1 Y2 (s)
This means that agents from any country n will be content to hold the available country mutual funds at prices V1m
S n X Y1w u0 [C2 (s)] m Y2 (s) = = (s) (s) Y m (s), m = 1, . . ., N. n u0 (C1 ) Y2w (s) 2 s=1 s=1 S X
"
At these prices,
N X
V1m
m=1
m=1
N X
( S X
Y1w (s) Y m (s) Y2w (s) 2 s=1 53
"
Y1w = (s) Y2w (s) s=1 = Y1w ,
S X
"
#" N X
Y2m (s)
m=1
#)
so indeed, condition (4) is satised. It remains only to calculate the riskless rate of interest. That comes from the Euler equation,
S S n X 1 u0 [C2 (s)] X C n (s) (s) n 1 . = = n 1 + r s=1 u0 (C1 ) C2 (s) s=1
But the consumption functions, together with (4), imply that for any country n,
n C1 N X m Y1n + V1n Y1w + V1 PN Y1w + m=1 V1m m=1 ! Y1n + V1n = Y1w = n Y1w , P Y1w + N V1m m=1
1 = 1+
Y1n + V1n = Y2w (s) = n Y2w (s), PN w m Y1 + m=1 V1 allowing us to express the equilibrium real interest rate in terms of exogenous variables through S X 1 Y w (s) w 1 . = 1 + r s=1 Y2 (s)
n C2 (s)
5. (a) With exponential utility the individual Euler equation for state s is exp(C1 ) = or, taking logs, (1 + r)(s) exp [C2 (s)] , p(s)
" #
(1 + r)(s) 1 C1 = C2 (s) log . p(s) Summing over the two countries implies Y1w = Y2w (s) (1 + r)(s) 2 log , p(s) 54
" #
which can be solved for p(s) = (s) exp [Y2w (s) Y1w ] . 1+r 2 Summing over states yields the (gross) interest rate, 1+r =
P
(5)
w s (s) exp 2 Y2 (s)
exp Y1w 2
h
i,
from which p(s) is easily calculated. (b) Notice that under the proposed consumption allocation markets clear on each date/state and, given the Arrow-Debreu prices calculated above, the complete-markets intertemporal Euler equations hold. For example, exp(C1 ) = exp Y1w exp() 2 w w w = exp [Y (s) Y1 ] exp Y2 (s) exp() 2 2 2 (1 + r)(s) exp [C2 (s)] , = p(s) where we have used (5) above. This shows eciency. It is also easy to check (it is a special case of part c below) that the Euler equations for equity shares hold at the implied equilibrium values of V1 and V1 (which also are given in part c). Now we check that the allocation satises budget constraints. Homes budget under the proposed equilibrium are 1 1 Y1 + V1 = V1 + V1 + B2 + C1 , 2 2 whereas Foreigns are 1 1 Y1 + V1 = V1 + V1 B2 + C1 , 2 2 If we substitute 1 1 C2 (s) = Y2 (s) + Y2 (s) (1 + r)B2 . 2 2 1 1 C2 (s) = Y2 (s) + Y2 (s) + (1 + r)B2 , 2 2

1 C1 = (Y1 + Y1 ) 2 55
into the rst Home budget constraint, we get 1 1 = B2 (Y1 Y1 ) (V1 V1 ). 2 2 1 [Y2 (s) + Y2 (s)] 2 into the second-period Home constraint gives C2 (s) = = (1 + r)B2 . (We could have gotten the same answers using Foreigns constraint, which means that Foreigns constraint holds once we nd and B2 such that Homes does.) Solving for and B2 yields 1 1+ 1+r

Substituting
1 1 = (Y1 Y1 ) + (V1 V1 ), 2 2
which has the interpretation that the present discounted value of the excess of Foreigns consumption over world average consumption equals the dierence between its and world average (equilibrium) date 1 resources. (c) If 6= , a natural conjecture is that the less risk averse (lower gamma) country holds a greater share of the risky world output portfolio. Thus, for all dates/states, one might conjecture that for some , C= Y w , + C = Y w + , +
which it will if
by analogy with the answer to part b. To support this (plainly outputmarket-clearing) allocation, V1 , say, would have to satisfy the Home Euler equation X exp(C1 )V1 = (s) Y2 (s) exp [C2 (s)] ,
s
V1 =
X
s
(s) Y2 (s) exp [Y w (s) Y1w ] . + 2 56
Given this price, Foreigns Euler equation for V1 is likewise satised at the candidate consumption allocation, as you can check. The same argument shows that V1 =
X
s
(s) Y2 (s) exp [Y w (s) Y1w ] + 2
is consistent with the Home and Foreign Euler equations at the proposed allocation. (Set = here to nd the prices relevant for part b above.) To check budget constraints, let /( + ) be Homes risky portfolio share. Then calculations analogous to those in part b show that and B2 satisfy Homes constraints (as well as Foreigns, by Walrass law) if = B2 [(1 ) (Y1 + V1 ) (Y1 + V1 )] , = (1 + r)B2 , which gives the solution for as 1 1+ 1+r

(Y1 + V1 ) (Y1 + V1 ) . + +
Here, depends not only on relative date 1 wealth but also on risk aversion. To see how, write the preceding as 1 1+ 1+r

[(Y1 + V1 ) (Y1 + V1 )] + (Y1 + V1 ) . + +
If the two countries have equal initial wealths, for example, the more risk averse country will have a higher deterministic consumption component. 6. (There is a mistake in the statement of the exercise. Delete the words each period in the third line from the bottom of the rst paragraph.) (a) Let us assume initially that risk-free bonds are indexed to the Home good (good X). (Part d below will consider the introduction of bonds indexed to
57
good Y.) In general, the period-by-period nance constraint for the Homecountry representative individual would be Cx,s + ps Cy,s + xx,s+1 Vx,s + xy,s+1 Vy,s + Bs+1 = (1 + rs )Bs + xx,s (Xs + Vx,s ) + xy,s (ps Ys + Vy,s ), Here xx,s and xy,s denote the fractional shares of the Home and Foreign country funds that the Home representative agent buys on date s 1, and rs is the risk-free own-rate of interest on good X between dates s 1 and s. (Remember that Vy,s is the ex dividend date s value of the Foreign country fund measured in units of good X.) The preceding constraint looks exactly like eq. (56) in the chapter, except that we recognize the distinctness of the Home and Foreign outputs and use the relative price of Y in terms of X on date s, ps, to express the budget constraint in terms of X. To nd a representative Home agents rst-order conditions, we use the Lagrangian approach (see supplement A to Chapter 2). Form the Lagrangian: Lt = Et
nX
s=t
st [u(Cx,s , Cy,s ) s (Cx,s + ps Cy,s + xx,s+1 Vx,s + xy,s+1 Vy,s
+ Bs+1 (1 + rs )Bs xx,s (Xs + Vx,s ) xy,s (ps Ys + Vy,s ))]} . Dierentiating with respect to xx,t+1 gives the rst-order condition for the Home country fund, t Vx,t = Et {t+1 (Xt+1 + Vx,t+1 )} . (6)
(Remember that xx,t+1 is a date t choice variable and that t and Vx,t are known to the consumer when that choice is made.) Similarly, dierentiating with respect to xy,t+1 yields the optimality condition for the Foreign country fund, t Vy,t = Et {t+1 (pt+1 Yt+1 + Vy,t+1 )} . (7) The condition for the riskless bond Bt+1 is similarly derived as t = Et {t+1 (1 + rt+1 )} . 58
Finally, we can determine t by dierentiation with respect to Cx,s and Cy,s , which reveals that the Lagrange multiplier is simply the marginal utility of the Home goods: u(Cx,t , Cy,t ) 1 u(Cx,t , Cy,t ) = t = . X pt Y (8)
(b) If Home and Foreign agents start with perfectly pooled, identical portfolios of risky claims, they will keep these portfolios and equal wealth levels forever. The reason is that under the initial allocation assumed, Home and Foreign agents are identical, not only ex ante but also ex post. Thus, unexpected shocks aect them equally, open up no opportunities for trade, and do not redistribute wealth between them. [If one did not start from such a perfectly pooled equilibrium, there would be no guarantee of reaching it, of course. The Lucas (1982) model is silent on how this assumed equilibrium is reached, whereas the model in section 5.3 does not require initial perfect pooling.] Equation (8) above shows that in the perfectly pooled equilibrium, u pt = u

1 X , 1Y 2 t 2 t 1 X , 1Y 2 t 2 t
(c) Applying iterative forward substitution to eqs. (6) and (7) above, coupled with a no speculative bubbles condition, we derive Vx,t = Et
X
/Y . /X
In addition,
s=t+1 X
st
u u
1 X , 1Y 2 s 2 s 1 X , 1Y 2 t 2 t
/X /X
Xs
st Vy,t = Et s=t+1 = pt Et
X s=t+1
u u
st
u u
1 X , 1Y 2 s 2 s 1 X , 1Y 2 t 2 t
1 X , 1Y 2 s 2 s 1 X , 1Y 2 t 2 t
/X /X

/Y /Y
ps Ys Ys
59
where the two alternative ways of expressing Vy,t are derived using the expression for p from part b. (d) An Euler-equation argument shows that the equilibrium date t price of a unit of X to be delivered with certainty on date t + 1, px,t , is px,t =
u 1 Xt+1 , 1 Yt+1 /X 2 2 Et . 1 1
Xt , 2 Yt /X
The Euler equation for the risk-free bond, derived in part a above, shows that 1 . px,t = 1 + rt+1 Analogously, the date t price (in terms of good X) of a unit of Y to be delivered with certainty on date t + 1, py,t , is py,t =
u 1 Xt+1 , 1 Yt+1 /X 2 2 Et 1 1
X , Y /X 2 t 2 t
The price of the same security in terms good Y on date t is py,t = pt Notice that the equation
u 1 Xt+1 , 1 Yt+1 /Y 2 2 Et u 1 X , 1 Y /Y
2 t 2 t
pt+1 .

py,t 1 = y pt 1 + rt+1
denes the own rate of interest on good Y between dates t and t+1. If we had explicitly introduced risk-free bonds denominated in good Y into the budget constraint of part a, we would have found the additional Euler equation for those bonds,
u (Cx,t , Cy,t ) u (Cx,t+1 , Cy,t+1 ) y , = 1 + rt+1 Et Y Y ( )
60
which is equivalent to u (Cx,t , Cy,t ) = Et X

y 1 + rt+1 pt+1
pt
This equation merely establishes the relationship between the own-rates of interest on the two goods; nothing in our analysis is changed, since bond markets actually are redundant in this model (they are never used in equilibrium). Notice that a more plausible choice for the riskless bond might be one with a face value indexed in some way to utility. When the period utility function takes the form u(Cx , Cy ) = u [(Cx , Cy )] = u(C), as in Chapter 4 [with (Cx , Cy ) homogeneous of degree one], then we can dene a bond that is indexed to real consumption C. The rate of interest on that bond denes the own rate of interest on the real consumption basket C, according to the Euler equation
c u0 (Ct ) = (1 + rt+1 )Et {0 (Ct+1 )} . u
u (Cx,t+1 , Cy,t+1 ) . X
In equilibrium, that interest rate is given by Et u0 1 Xt+1 , 1 Yt+1 1 2 2 h i = c 0 1X , 1Y 1 + rt+1 u t 2 t 2

n h io
61
Workbook2
Chapter 6 Solutions
1. (a) We look at the symmetric ecient incentive-compatible contract. That contract maximizes an equally-weighted average of Home and Foreign expected utility, E {u(C)} + E {u(C )} , subject to the constraints Y P () Y, which must hold for all N possible realizations of the shock . The Lagrangian for the contracting problem is L = max
P () N X
1=1 N X 1=1
N X + i ) (i ) P (i ) (Y (i ) P (i ) (Y i ) . 1=1
(i ) u Y + i P (i ) + u Y i + P (i )
The rst-order condition with respect to P (i ) is (i ) {u0 [C(i )] + u0 [C (i )]} (i ) + (i ) = 0, and the complementary slackness conditions are (i ) (Y + i ) P (i ) = 0,
1
(1)
62
Despite the forbidding formalism of the Kuhn-Tucker conditions, the solution to the problem can be characterized rather simply. Consider the solution in states where the incentive constraints do not bind, that is when (i ) = (i ) = 0. Across these states, the rst-order condition (1) reduces to u0 [C(i )] = u0 [C (i )], and so C(i ) = C (i ). There is therefore a range [e, e] such that inside this interval C = C . The bound e is easily found as the largest such that Y + , implying e= Y. 1 (2) (b) For > e, the incentive constraint P (i ) (Y + i ) prevents full insurance, so P (i ) = e + (i e) [where we have substituted for Y from equation (2)]. Since the equilibrium is symmetric, for < e, P (i ) = e + (i + e). To graph the payments schedule that the contract implies, put P () on the vertical axis and on the horizontal axis. The payments schedule passes through the origin, has slope 1 over [e, e], and has slope outside that interval. See gure 6.1. 2. This problem is completely parallel to the problem in the text. There are only two dierences: the zero-prot condition for lenders is now
N X 1=1
(i ) (Y i ) + P (i ) = 0.
(i )P (i ) = (1 + r)D,
and P (i ) 0 must obtain (because the country, by assumption, cannot receive any payments in period 2). It is easy to see that the country will not in general be able to attain as much insurance in this environment as in the two-way payment environment of the text. The basic problem is that the amounts lenders must collect are larger here because the countrys net second-period payments must be positive. (The countrys second-period 63
income is also higher as a result of prior investment, but as long as < 1, the former eect dominates.) For example, in the case of pure insurance contracts (as in the text), the necessary condition for full insurance to be feasible is (Y + ) or Y. 1
Here, however, the corresponding condition is + (1 + r)D Y + + (1 + r)D
[because the contract with P () = + (1 + r)D, which yields full insurance, satises the zero-prot condition, and satises P () 0, , is incentivecompatible only if the preceding inequality holds]. The preceding inequality is equivalent to + (1 + r)D Y. (3) 1
Since the minimum amount the country needs to borrow to achieve full insurance while keeping its second-period payments nonnegative is /(1+r) > 0, full insurance therefore is feasible if and only if Y. 1
Clearly, given that > 0, the constraint under pure insurance contracts is less likely to be binding than in the case where insurance is achieved through an output-indexed borrowing contract. [Compare eq. (3) with the last term in footnote 8 of the chapter.] (a) The reference to a given D in the exercise obviously refers to a level D small enough that the country can be compelled to make nonnegative payments with an expected value of (1 + r)Dcreditors would never agree to lend a larger amount in the rst place! Qualitatively, the equilibrium is 64
the same as in section 6.1.1 of the text. Over a range of below a cuto value e, dP ()/d = 1, while above e, dP ()/d = . For the optimal level of D, which now is generally smaller than D = /(1 + r), P () = 0. The optimal choice of D minimizes the expected value of the repayments the country must make. (b) The country is clearly worse o if it only has access only to equity contracts, since in eect it must be able to commit to repayments with a positive expected value instead of repayments with a zero expected value. 3. Consumption in each period is given by Ct = F (Dt ) + t P (t ), (4)
where P (t ) 0 and Dt 0. The nonnegativity constraint on P (t ) reects the assumption that it is not feasible to write and to enforce contracts that require insurers to indemnify the sovereign after the realization of a bad state of the world. The nonnegativity constraint on Dt reects the assumption that the country cannot lend abroad. Since we assume a competitive lending market, the risk-neutral foreign lenders receive an expected return of 1 + r in equilibrium, that is, X e (t )Pt1 (t ) = (1 + r)Dt , (5)
t e where Pt1 (t ) is the amount of debt servicing that the lenders in period t 1 expect to receive in period t as a function of the realization t .
(a) Suppose that the country can commit itself in period t 1 to a payment schedule in period t, given by Pt1 (t ). In that case the country would determine the lenders expectations of the actual level of debt servicing:
e Pt1 (t ) = Pt1 (t ).
(6)
Because the analysis assumes t to be stationary, the countrys choices are time invariant. The optimal P (t ) and D can be computed by maximizing 65
Et1 u (Ct ) subject to the constraints given by eqs. (4), (5), and (6). The Lagrangian for the countrys problem is: " # X X L= (t )u [F (D) + t P (t )] + (t ) P (t ) (1 + r) D .
t t
Taking derivatives with respect to P (t ) and D we obtain u0 [C(t )] = for all t and F 0 (D) = 1 + r. Combining the preceding two equations with (4) and (5), we obtain the critical values for P (t ) and D: e D = max D, , (7) 1+r P (t ) = t e + (1 + r)D . (8) Equations (7) and (8) indicate that the country, by irrevocably committing not to repudiate, is able to invest eciently, thereby maximizing expected consumption. It also achieves ecient risk shifting. Equation (8) gives the debt-servicing commitment P (t ) . It calls for adding the dierence, which can be positive or negative, between the realization of t and e = Et1 {t } to repayment of loans D at the interest rate r. Equation (7) gives an amount of borrowing that allows investment up to a point where the marginal product of capital equals 1+r, yet is also sucient for lenders to prepay the indemnity associated with the worst possible state of the world (equal to the discounted value of the dierence between e and ). Why? Consumption is independent of the state of the world; by eq. (4) it is stabilized at Ct = C = F (D) (1 + r)D + e (for all t). But if D is less than (e )/(1 + r), ecient risk shifting with nonnegative country payments to creditors requires borrowing beyond the 66
minimum level D necessary for ecient investment. [Recall that for D > D, F (D) (1 + r)D = F (D ) (1 + r)D , so consumption remains at C for D > D.] We next consider the conditions under which it is possible to sustain the above contract as a trigger-strategy equilibrium where the only penalty to default is that the country is excluded from all future borrowing. Suppose that on date t the country considers default. Its short-run gain is the extra utility on date t from avoiding repayment, Gain(t ) = u [F (D ) + t ] u C .
X st
The date t cost associated with default is Cost =
s=t+1
u(C) Et
s=t+1
st
u(t ) .
Since the economy is stationary we can drop the time subscript and rewrite the cost as u(C) Eu() . Cost = 1 Since the utility function is strictly concave and since domestic investment is protable, u(C) > u(E) > Eu(). There is therefore a positive cost to default. The commitment contract is sustainable in all states of nature (and on all dates) only if Gain() Cost. that is, when u [F (D ) + ] u C u(C) Eu() . 1
[For more details and analysis of the case where the parameters of the model are such that trigger-strategy expectations cannot support the ecient allocation, refer to Herschel I. Grossman and John B. Van Huyck, Sovereign
67
debt as a contingent claim: Excusable default, repudiation, and reputation, American Economic Review 78 (December 1988): 1088-97.] (b) Recall the discussion of section 6.2.2. in the text. (c) Recall the discussion of section 6.1.2.4 in the text. 4. (a) If E2 is very largespecically, if E2 (1 + r)(I Y1 ), where I is the rst-best ecient investment level dened by eq. (44) in Chapter 6then entrepreneurs can nance their investments with noncontingent loans such that P (Z) = P (0), and the moral hazard problem disappears. So assume that E2 < (1 + r)(I Y1 ). With observable second-period endowment E2 , the agents maximization problem becomes EC2 = (I) [Z P (Z)] [1 (I)] P (0) + (1 + r)L + E2 = (I) [Z P (Z)] [1 (I)] P (0) + (1 + r)(Y1 + D I) + E2 , and the bankruptcy constraint becomes P (0) E2 . By the same logic as in the text, this constraint must hold with equalityif it did not, the gap between P (Z) and P (0) would be unnecessarily large, suboptimally distorting investment. Given these modications, the incentive compatibility condition becomes 0 (I) {Z [P (Z) E2 ]} = 1 + r, implying that the IC curve of the text is given by P (Z) E2 = Z 1+r . 0 (I)
The ZP (zero prots for lenders) curve is now (I)P (Z) + [1 (I)] E2 = (1 + r)D = (1 + r)(I Y1 ) 68
or
(1 + r)[I Y1 E2 /(1 + r)] . (I) Thus the IC and ZP curves are analogous to those of the text, except that P (Z) is replaced by P (Z) E2 and Y1 is replaced by presented discounted collateralizable income Y1 + E2 /(1 + r). Here, investment I responds to Y1 + E2 /(1 + r) in the same way as to changes in Y1 . P (Z) E2 = (b) Now assume that the government has a debt outstanding to foreigners of Dg per capita, payable in the second period. The government, of course, can only impose second-period taxes on successful entrepreneurs. The governments budget constraint is given by (I) = Dg ,
where is the tax on successful entrepreneurs. (Note that since the returns on various individual projects are independent, there is no uncertainty in the aggregate.) With this tax the IC curve becomes 0 (I) {Z [P (Z) + ]} = 1 + r or P (Z) + = Z 1+r , 0 (I)
while the ZP curve remains the same as eq. (49) in Chapter 6. However, adding to both sides of the ZP curve and making use of the government budget constraint implies that the ZP can be written as P (Z) + = (1 + r)[I Y1 + Dg /(1 + r)] . (I)
A rise in Dg therefore has an eect exactly analogous to a fall in E2 in part a. The overhang of government debt discourages investment. 5. First, note that in the decentralized economy, the presence of savers has no eect since the entrepreneurs already face an innitely elastic supply of 69
world savings. One can then show that a planner who has no information advantage over the private sector cannot achieve a Pareto improvement. One way to approach the problem is to have the planner maximize the utility of the representative entrepreneur subject to the constraint that savers earn at least the world market rate of return, that is, that (I)P (Z) + [1 (I)] P (0) (1 + r)D. This problem can be solved as in the text, in which case the same algorithm shows that the above constraint is binding. (If any multiplier in the problem is strictly positive, they all must be.) To obtain a more intuitive understanding of the preceding argument, let us consider the tax redistribution scheme suggested in the problem. In the rst-period, the government places a tax on savers of 1 , and gives the proceeds to entrepreneurs. (Obviously, other things equal, this will raise total output though savers will be worse o.) Then, in the second period, the government issues a tax on (successful) entrepreneurs of 2 , and uses the proceeds to pay back savers. If savers are to be made no worse o, then we must have 2 (I) (1 + r)1 . We assume that there are an equal number of investors and savers. In the presence of transfers, the entrepreneur maximizes EC2 = (I) [Z P (Z) 2 ] [1 (I)] [P (0)] + (1 + r) (Y1 + 1 + D I) . The incentive compatibility constraint is now 0 (I) {Z [P (Z) + 2 P (0)]} = 1 + r, which, with P (0) = 0 (an unsuccessful entrepreneur does not pay the tax), reduces to 1+r (9) P (Z) + 2 = Z 0 . (I) 70
Since entrepreneurs now borrow D = I Y1 1 , the zero-prot condition is (1 + r) (I Y1 1 ) = (I)P (Z) + [1 (I)] P (0). If we substitute the condition 2 (I) = (1 + r)1 (which should hold so as not to make the savers worse o), the zero-prot condition can be written as P (Z) + 2 = (1 + r)(I Y1 ) . (I) (10)
It follows clearly from equations (9) and (10) that the IC and ZP curves are unchanged, except that P (Z) + 2 has replaced P (Z) on the vertical axis in gure 6.11 of the text. Each entrepreneur borrows 1 less, and the successful ones pay (1 + r)1 /(I) = 2 less to creditors. The intervention therefore has no eect on investment. The logic here is the same as in exercise 4, part b. Each entrepreneurs rst-period income rises by 1 . But the government must raise second-period taxes on successful entrepreneurs by an amount sucient to raise revenue (1 + r)1 . This overhang eect exactly osets the gain from the rst-period subsidy. In general, there is no scope for Pareto-improving government intervention here despite the credit market imperfection. 6. (a) Assuming there is no debt forgiveness, the countrys maximization problem can be written as max {log C1 + log C2 }
I
subject to C1 = Y1 I, C2 = (1 )I . (By denition, the inherited debt specied in this exercise yields no benets in period 1.) The maximization problem can be rewritten as max {log (Y1 I) + log [(1 )I ]} .
I
71
The rst-order condition with respect to I is 1 1 = , (1 )I 1 = Y1 I (1 )I I Id = Y1 1 +
giving an investment level of
when default is planned. The countrys repayment is then the (forcibly extracted) amount Y1 d R = . 1 + The equilibrium is at point A in gure 6.2 (cf. gure 6.3 on p. 382 in the chapter). Note that the parameter , which eectively determines the productivity of investment, does not aect the optimal level of investment. This is the result of our log utility specication which implies that income and substitution eects exactly cancel and thus that the tax rate does not aect savings. (b) Now assume that creditors agree to write down the countrys very large debt to Rd . That is, once the country has paid Rd , it does not owe anything further. Why does this aect the countrys maximization problem? Intuitively, once the country pays back its debt, the marginal tax on investment drops to zero. Y1 d With the debt written down to R = , the period 2 budget 1 + constraint becomes Y1 C2 = max (1 )Y2 , Y2 . 1 + It is easiest to proceed by assuming that the country is going to pay back in full; it is easy to show that after forgiveness to Rd , this strategy dominates any strategy involving default. The countrys maximization problem now becomes max {log C1 + log [(Y1 C1 ) Rd ]}
C1
72
and the rst-order condition becomes 1 = (Y1 C1 )1 = > , d (Y C )1 C1 C2 Y1 C1 R 1 Y1 C1 1 from which it immediately follows that C1 must be lower with debt forgiveness than its level C1 = Y1 /(1 + ) in part a. Correspondingly, investment must be higher. (Intuitively, in comparison to part a, two things happen: falls to zero, which leaves investment unchanged under log preferences, but a second-period lump-sum levy of Rd is imposed, reducing consumption and raising investment.) The country must also be better o, since with debt forgiveness to Rd , it always has the option of replicating its allocation in the no-forgiveness equilibrium. We can also see this graphically in gure 6.3. Figure 6.3 is similar to the graph on p. 382 in the text. The line GDP represents the production possibilities frontier for the small country. GNPd represents the budget constraint for the country in the case where the debt is not reduced and repayment of Y2 is enforced. As discussed in part a of this problem the equilibrium for the economy is at A. GNPn represents the budget constraint for the economy when the debt is reduced to Rd and the country does not default. Clearly GNPn intersects GNPd at point A and has a steeper slope at that point. There is therefore a point to the left of A on the GNPn curve (A0 ) that the small country strictly prefers to point A. This ranking can be seen from the tangencies of the indierence curves. (c) In gure 6.3 we see that by writing the debt down to the level Rd dened in part b, lenders can raise borrower investment and welfare while averting default and leaving their own repayment unchanged at (I d ) . At point A0 in gure 6.3, however, the borrower strictly prefers repayment to default. Thus, lenders could forgive slightly less of the countrys debt, shifting GNPn downward and inducing the country to invest and repay even more than the amounts shown in gure 6.3. The countrys optimal investment given a binding enforcement constraint is I d = Y1 /(1 + ), which yields a utility
73
level under default of Y1 U = log 1 +

d
Y1 . + log (1 ) 1 +
Selsh creditors would write the debt down by the smallest amount needed to dissuade the country from defaulting (and thereby obtaining utility U d ). The problem for the creditors thus can be written as max R subject to 0= and I I Y1 I I R (11)
U d log(Y1 I) + log (I R) , where condition (11) states that for a given level of remaining debt R after forgiveness, investment is chosen optimally by the indebted country, given that it repays R in full. The optimal level of debt after forgiveness, R , is represented in gure 6.4. GNPr corresponds to the nondefault budget constraint when R = R . (It is simply the curve GNPn corresponding to the repayment-maximizing level of debt forgiveness.) Clearly, the optimal R is greater than Rd . The optimal strategy for creditors is to lower debt to the point where the debtor is just indierent between its optimal strategies under repayment and under default. The optimal consumption point for the country is B (as drawn in gure 6.4). Investment I r is plainly greater than I d.
74
Workbook2
Chapter 7 Solutions
1. (The word equiproportionate in the third line of the statement of this exercise should be lump-sum.) (a) With the introduction of tax-nanced government spending in the Weil (1989a) model, the period budget constraint for a family of vintage is given by kt+1 = (1 + rt ) kt + wt c t (where is the lump-sum tax) instead of by eq. (30) in Chapter 7. We write the above expression in average per capita terms as kt+1 kt = nkt f (kt ) ct g , 1+n 1+n (1)
giving the analog of eq. (32) in the chapter. Here we have used the balancedbudget constraint g = . The introduction of government expenditure therefore shifts the k = 0 locus down by g/(1 + n) in the phase diagram for per capita consumption and the capital-labor ratio (gure 7.7 on p. 449). The presence of tax-nanced government spending does not aect eq. (33) in the text: ct+1 = [1 + f 0 (kt+1 )] [ct n(1 )kt+1 ] . (2)
1
75
(b) Figure 7.1 shows the phase diagram. In that gure, an unanticipated permanent rise in g shifts the k = 0 schedule down. At the instant the shock occurs, consumption declines immediately from point A to point B. In subsequent periods, there is a gradual decumulation of capital, and consumption continues to decline until the new steady state A0 is reached at lower levels of c and k. The decline in k represents a crowding out eect of balanced-budget government spending. (c) As can be seen in gure 7.2, the impact eect of the announcement is an immediate decline in consumption from c0 to c1 . The economy then moves along an unstable path relative to the initial steady state. Along that path, c gradually declines while k increases until the economy reaches point D on date T . Point D lies on the stable path corresponding to the new constant level of g. After date T , c and k both decline until the new steady-state cnew , knew is reached (point A0 ). Per capita consumption and the capital labor ratio are lower in the new steady state relative to the steady state with g = 0. 2. (Note that there is an obvious typo in the question. The production function is Ak and not Ak.) The economys initial equilibrium is the autarky steady state as in the discussion in section 7.2.2.3 in the book. The exercise states that in that steady state the autarky interest rate ra equals the world interest rate r. (In particular, this means that initially the economy is not borrowing-constrained.) For an arbitrary productivity parameter A in the production function, the autarky steady-state capital stock is given by (1 )A (ka ) wa ka = = 1+ 1+ or a A(1 ) 1 k = . 1+ Using the last expression to substitute for ka , we see that the equality of the 76
" #
1
world and steady-state autarky interest rates is written as ra = A k a =

1
(1 + ) 1 (1 ) = r,
1 (3)
where we recall that the rate of depreciation, , is 100 percent. Notice that the second and third equalities above are independent of the parameter A. This means that after A rises permanently from A = 1, the economy necessarily returns to a new unconstrained steady state with ra = rd = r, but with a higher capital stock. Another way to see that result is to note that the equality w ku 1+ ensures convergence to the world interest rate [cf. eq. (60) in Chapter 7; w is the unconstrained steady-state wage]. But it us easy to check that a rise in A from A = 1 simply multiplies both w and ku above by the same factor A1/(1) , leaving the preceding inequality intact if it also held before the rise in productivity. The economy can move to its new steady state in a single period only if w0 A + w0 1+ 1+r
1 1
(4)
where the right-hand side is the new steady-state capital stock and w0 is the wage in the rst period A rises, given the predetermined capital stock: (1 ) w0 = (1 )A 1+ Using eq. (3), we may express inequality (4) as
" #
1
(1 ) + (1 )A 1+ 1+ 77
"
A(1 ) 1+
"
1 1
which is equivalent to
+ A 1 1+
. 1+
This inequality is quite intuitive. If the productivity increase is very small, the economy will reach its new (unconstrained) steady state in a single period with even a small amount of loans from abroad ( small). If, in contrast, A rises by a very large amount, convergence will be slower, because in the short run wages rise only by a factor of A, whereas the long-run unconstrained capital stock rises by the bigger factor A1/(1) . (The ratio of the two is the term A/(1) that appears in the preceding inequality.) 3. The planners problem is to maximize Ut =
X s=t
st log Cs
(5)
subject to the production technology in the R&D sector, At+1 At = At LA,t , labor-market clearing, L = LA + LY , the nal-goods production function, Yt = L1 At Kt , Y,t and the social budget constraint Yt = Ct + At+1 Kt+1 . The Lagrangian for the maximization problem is L=
X s=t st log[As Ks (LLA,s )1 As+1 Ks+1 ] + s (As+1 As As LA,s )
(6)
(7)
(8)
(9)
78
(see footnote 42 on p. 491). The previous expression follows from using (6), (7), (8), and (9) to substitute for C in (5). Next we take derivatives with respect to LA,t , Kt+1 and At+1 : LA,t : At+1 :
"
At Kt ( 1)(L LA,t ) = t At , Ct
Kt+1 (L LA,t+1 )1 Kt+1 + t = t+1 (1 + LA,t+1 ) Ct Ct+1
"
#)
Kt+1 :
At+1 Ct
1 At+1 (L LA,t+1 )1 Kt+1
Ct+1
We are looking for a steady-state equilibrium with constant real interest rate, constant K, constant relative prices, and a constant allocation of labor across the two sectors; that is, we are looking for a balanced growth path. Output and consumption will grow at a rate g in steady state. Along a steady-state path the preceding rst-order condition with respect to Kt+1 can be written more simply as: h i 1 + g = (L LA )1 K 1 . (10) The rst-order condition with respect to LA,t simplies to: t = ( 1)K (L LA ) . Ct (11)
Substitute (11) into the rst-order condition for At+1 (with Kt+1 = K in steady state). Then multiply through by Ct+1 to obtain (1 + g)( 1)K (L LA ) (12) h i ( 1) (1 + LA )K (L LA ) K (L LA )1 . = (1 + g)K + Also, At+1 At = LA = g. At 79 (13)
[Equation (13) follows from (6).] Along a balanced growth path, the number of capital good types grows at rate g, whereas the quantity of each type of capital good remains constant at K. Finally, using (10), (12), and (13), we can solve for g, K, and LA : g = L (1 ), K=
1 1
1 (1 + L) 1 ,
LA = L
(1 + g)(1 ) .
[Hint: To solve for g, divide (12) through by K (L LA )1 , then use (10) and (13).] As expected, the growth rate for the planners problem is unambiguously higher than in the laissez-faire equilibrium. This discrepancy arises because in the decentralized solution there are two distortions. First, rms in the R&D sector do not internalize the fact that their inventions will lower the cost of producing future inventions. Second, for any given allocation of labor, monopolistic suppliers set K lower than the planner would. That is, they underutilize inventions, creating a static ineciency (albeit one that aects R&D employment). 4. In section 7.4.1, we saw that for logarithmic utility equilibrium consumption is Ct = (1 )At Kt . We simply check that the consumption function in the question also holds true. Note that ws = (1 ) As Ks and rt = At Kt1 1 (due to the assumption of 100 percent capital depreciation in the rst period of use). Also valid under the preceding solution for consumption is the equality u0 (Cs ) At Kt Ct = = u0 (Ct ) Cs As Ks
80
for all t and s, implying that

X s=t
Et
st u0 (Cs ) ws L u0 (Ct )
= (1 )At Kt 1 At Kt . = 1
X s=t
st
(Recall that L was normalized to equal 1 in section 7.4.1.) Simple substitution therefore yields the required result, Ct = (1 )
At Kt
1 At Kt = (1 )At Kt . + 1
Interpretation: Under log utility optimal consumption is a fraction (1 ) of lifetime wealth; see supplement A to Chapter 5. The latter, in turn, is the sum of nonhuman wealth (1 + rt )Kt and human wealth
X s=t
Et
st u0 (Cs ) ws L u0 (Ct )
(the present discounted value of current and future labor earnings).
81
Workbook2
Chapter 8 Solutions
1. First think about the continuous-time case. At time t = 0 the market believes the governments promise that it will peg the exchange rate at its t = T level from time T onward. In that case, we can show that the exchange rate is indeterminate, so that the governments policy is not coherent (in that it does not tie down a unique market equilibrium). Let perfect-foresight equilibrium be described by the continuous-time Cagan model [eq. (70) on p. 559 of the book], mt = et et . Let ea be an arbitrary time-T exchange rate and suppose the market rmly T expects that rate to prevail. Then the preceding Cagan equation, coupled with the terminal condition eT = ea , shows that the exchange rate path T et = 1Z T exp[(t s)/]ms ds + exp[(t T )/]ea T t
will equilibrate markets for t [0, T ]. Next consider the monetary authoritys position at time T when confronted with this exchange rate path. The authority has no choice, in view of its vow of a constant exchange rate from T on, but to set the fundamental mT = ea and to hold mt = ea for all t > T . Why? Were the authority to T T act otherwise, the exchange rate would deviate from ea at some point in the T
1
82
interval [T, ), in violation of the governments initial pledge. In short, the authority must validate (ratify) any market expectation whatsoever. Any initial exchange rate can be an equilibrium because each is conditioned on a dierent expectation of what the (constant) money supply path will be from date T onward. The situation is subtly dierent in discrete time, in which case the model is mt = et (et+1 et ) . The basic reason discrete time makes a dierence is that now, if the market rmly foresees a date T rate of ea , there are two distinct ways the authorT ity can fulll its promise of a constant exchange rate from date T on (two alternatives which collapse to one in continuous time). First, the authority could still set mt = ea for t = T and for t > T , as in the continuous-time T settingin which case the exchange rate again is not uniquely determined. Alternatively, if the authority can commit not to adjust mT fully to validate ea , but instead only to set mt = ea for t strictly greater than T only, we T T again get an exchange rate path constant at ea starting on date T . In this T second case, however, the exchange rate is uniquely determined. To x ideas, suppose the authority can commit to a xed value mT for date T . In that case, since the date T + 1 rate will be pegged at its date T level, the Cagan equation says that the exchange rate must satisfy eT = mT on date T , a well-dened solution. The distinction we make here actually does arise in practice. For example, the Maastricht Treaty on European Union (in eect) links to prior market values the irrevocably xed bilateral currency conversion factors that will apply to member currencies when European economic and monetary union (EMU) starts on 4 January 1999. (EMU starts ocially on Friday, 1 January 1999, but the rst business day for the new European System of Central 83
Banks is Monday, 4 January.) Member currency bilateral rates must be xed forever at their 31 December 1998 levels. This requirement need not lead to exchange-rate indeterminacy, however, even if participating national central banks do not intervene to set bilateral currency values on the last day of 1998. The reason is that the national central banks that will still be in operation through the end of 1998 have no automatic incentive to ratify market exchange rates on the last day before EMU. For a full discussion, see Maurice Obstfeld, A strategy for launching the Euro, European Economic Review 42 (May 1998). 2. (a) Using the individuals budget constraint to substitute for Ct in the utility function, and taking derivatives with respect to Bs+1 and Ms , one obtains the following rst-order conditions: Bs+1 : Ms : u0 (Cs ) = (1 + r)u0 (Cs+1 ), Ms u0 (Cs ) 1 Y g0 Ps Ps

u0 (Cs+1 ) . Ps+1
Using the consumption Euler equation, we can rewrite the money demand equation as is+1 0 Ms = . Yg Ps 1 + is+1 (b) If one assumes there are no speculative bubbles, the price level grows at the same gross rate 1 + as the nominal money supply. Steady-state real money balances thus are M 1 = g0 1 P Y
"
1 1+
!#
84
(c) With the new budget constraint, the rst-order conditions become: Bs+1 : Ms u (Cs )g Ps
0
Ms+1 = (1 + r)u (Cs+1 )g , Ps+1

0
Ms :
Ms+1 = u (Cs+1 ) g Ps+1

0
Ms u0 (Cs ) g Ps
"
1 Ms Cs M g0 Ps g s Ps Ps
!
Ps+1
1 Ps
One can then rewrite the money demand equation as Ms is+1 P Cs s2 = . 1 + is+1 Ms g Ps g0 (d) The analysis here parallels that in the text. 3. (a) The constraint follows by straightforward addition. Because domestic money is a nontraded asset, the present value of private and government spending must equal the present value of the economys tradable resources, which in turn equals the sum of the present value of output and the economys net foreign nancial wealth. (b) The answer does not change. The variable Bt in part a, the overall net foreign assets of the economy as a whole, equals the sum of domestic g p government and private-sector net assets, Bt = Bt + Bt [recall eq. (7) from Chapter 3]. Equation (38) in Chapter 8 would change, however, in that p (1 + r)Bt rather than (1 + r)Bt would appear on its right-hand side. The last equation in footnote 26, p. 537 (the government budget constraint) would g also dier, in that (1 + r)Bt would be added to its right-hand side. (c) Let us take the setup of Chapter 4, but with the services of money being the nontraded good and with is+1 /(1 + is+1 ) that goods date s price in

85
terms of the tradable, consumption (recall section 8.3.3). When = 1 (the Cobb-Douglas case), footnote 22, Chapter 4, tells us that C, and that Psc =
M P
C (M/P )1 (1 )1
!1
is+1 1 + is+1
Equation (26) of Chapter 4, translated to apply to the current setting, is the Euler equation for real consumption,
Cs+1
Ms+1 Ps+1
!1
(1 + r)Psc = c Ps+1
!
"
Cs
Ms Ps
With = 1, we also have that Ms = Ps
1 (Psc )1/(1) Cs
[eq. (40) in Chapter 8]. Substituting this relation into the Euler equation preceding it, we derive Cs+1 =
Psc c Ps+1
!1
(1 + r) Cs ,
which parallels eq. (34) in Chapter 4 for the case = 1. If you combine this equation with the intertemporal constraint derived in part a of this exercise, the result is (1 + r)Bt +
P P 1 st (Y
s=t 1+r s
Ct =
1 ]st s=t [(1 + r)
The denition of the consumption-based real interest rate (section 8.3.3) leads to the equivalent formula Ct = (1 + r)Bt +
P hQs
s=t
c Pt c Ps
Gs )
c v=t+1 (1 + rv )
P 1 st (Y
s=t 1+r
i1
Gs )
(st)
86
When = 1, equilibrium consumption is simply Ct = (1 ) (1 + r)Bt + as in a model with no money. (d) In the more general case in which 6= 1, the consumption Euler equation is ! Ptc (1 + r) Ct , Ct+1 = c Pt+1 which follows directly from eq. (34) in Chapter 4. Solving as in part c leads to st P 1 (1 + r)Bt + 1+r (Ys Gs ) s=t . Ct = c P 1 ]st Pt c s=t [(1 + r) P
s
"
X s=t
1 1+r
st
(Ys Gs ) ,
4. For a time-varying real interest rate in eq. (59) of Chapter 8, the consumption Euler equation is Ps Ps1 0 u (Cs ) = (1 + rs ) u0 (Cs+1 ). Ps Ps+1 Making use of the Fisher parity equation 1 + is+1 = (1 + rs+1 ) (Ps+1 /Ps ) , we divide both sides by 1 + rs to derive u0 (Cs ) u0 (Cs+1 ) = (1 + rs+1 ) . 1 + is 1 + is+1
rather than 0, while it is still true that Et (dkt+h )2 hv2 (because terms multiplied by h2 and h3/2 disappear as h 0, being of order greater than h). Thus, hv 2 00 G (kt ) Et dG(kt+h ) = hG 0 (kt ) + 2 87
5. Under the revised fundamentals process, we still have eq. (83), p. 571, but now Et {G 0 (kt )dkt+h } = hG 0 (kt )
n o
v2 00 G (k) 2 as the dierential equation any exchange rate solution must satisfy [cf. eq. (84), p. 572]. A general solution is of the form G(k) = k + G 0 (k) + G(k) = k + + b1 exp(1 k) + b2 exp(2 k) where the bs are arbitrary constants. Because internal consistency requires that k + + b1 exp(1 k) + b2 exp(2 k) = k + [1 + 1 b1 exp(1 k) + 2 b2 exp(2 k)] +
i v 2 h (1 )2 b1 exp(1 k) + (2 )2 b2 exp(2 k) , 2
in the limit as h 0. Plugging this result into eq. (82) on p. 571 leads to
= and 1 and 2 are the two roots of the quadratic equation v2 2 + 1 = 0. 2 The particular target zone solution still satises the smooth pasting conditions G0 (k) = 0 at the top and bottom of the band. The argument is the same as in section 8.5.4, because at the edges of the zone Et {dkt+h } still changes discontinuously even when 6= 0 (movements that would drive the exchange rate out of the band suddenly are prohibited). 6. (a) Intuitively, spending a domestic currency unit on the least-cost consumer basket such that C = (C1 , ..., CN ) = 1 yields 1/P units of C. Alternatively, spending the currency unit on any individual commodity j yields 1/pj units of that good, each increasing C by j C/Cj . At an optimum these two uses of the currency unit must have the same impact on C, so that the equality 1 C 1 = P pj Cj 88
follows. Alternatively and more formally, consider the optimization problem P that denes the CPI P , which is to minimize the expenditure P = N pj Cj j=1 subject to (C1 , ..., CN ) = 1. One way to solve this problem is to set up the Lagrangian L=
j=1 N X
pj Cj [(C1 , ..., CN ) 1] .
The rst-order optimality conditions are (for all j): L = pj j = 0. Cj Since (C1 , ..., CN ) is linear homogeneous, we have
N X
j Cj = (C1 , ..., CN ) = 1
j=1
at the optimum. Thus, multiplying the preceding rst-order condition by Cj and summing over all j, we derive P =
N X
pj Cj =
j=1
j=1
N X
j Cj = .
This equality, however, allows us to write the rst-order condition as j = C pj = . Cj P
(b) The dierence between the ex post real return on a nominal Homecurrency bond and that on a nominal Foreign-currency bond is (1 + it+1 )Pt (1 + i )Pt t+1 , Pt+1 Pt+1 where P and P are the consumption-based price levels measured in Home and Foreign currency, respectively. Observe that these consumption-based price levels will be linked by purchasing power parity (absent trade barriers), 89
because they simply measure the price of a single basket of commodities in two currencies. Thus we may write the preceding real return dierential as (1 + it+1 )Pt (1 + i )Pt /Et t+1 . Pt+1 Pt+1 /Et+1 Equation (104) in Chapter 8 (covered interest parity) allows us to write the foregoing dierence as (1 + it+1 )Pt (1 + it+1 )Pt /Ft , Pt+1 Pt+1 /Et+1 and eq. (116) therefore implies that Et
(" # )
(1 + it+1 )Pt Et+1 (1 + it+1 )Pt /Ft u0 (Ct+1 ) Pt+1 Pt+1 u0 (Ct )
= 0.
Factoring out the term (1 + it+1 ) Pt /Ft , which is date t information, we are left with ( ! 0 ) Ft Et+1 u (Ct+1 ) 0 = Et . Pt+1 u0 (Ct )
(c) Substitute Et+1 Pt+1 = Pt+1 into the previous equation and multiply by 1/Ft to get 1 1 0 Ft Et+1 u (Ct+1 ) 0 = Et . 0 Pt+1 u (Ct )
Observe that this condition is perfectly symmetrical to the one derived in part b from the Home investors perspective. In the risk neutral case, the marginal utility of consumption is constant, and therefore we have that
1 Ft
Et
Pt+1
Et+1
= 0 = Et
Ft Et+1 . Pt+1
90
Plainly Siegels paradox does not apply. (d) The result follows immediately from part a, where it was shown that C/Cj = pj /P . (e) The condition that consumers equate their marginal rates of substitution to relative prices implies that Cx,t = Cy,t

1
!
Et p y,t , px,t
!
so that we can express the spot exchange rate as Et = 1 px,t Cx,t . p Cy,t y,t
In a perfectly pooled risk-sharing equilibrium, Cx,t = Cx,t = Xt /2 and Cy,t = Cy,t = Yt /2. Moreover, using the (binding) cash-in-advance constraints for the two currencies, Mt = px,t Xt and Mt = p Yt , we can express the equation y,t for the spot exchange rate as
Et = Using the result in part d:

(
Mt . Mt
)
Et+1 uj (Ct+1 ) Et pj,t+1 ( ) , j = x, y. Ft = uj (Ct+1 ) Et pj,t+1 In combination with the cash-in-advance constraints and the preceding relationship between the spot exchange rate and money supplies, we have, for j = x, ( ) Cx,t+1 ! Et ux (Ct+1 ) 1 Mt+1 ( ). Ft = Cx,t+1 Et ux (Ct+1 ) Mt+1 91
If money and output shocks have statistically independent distributions, however, we may factor out the terms in consumption above and write 1 ! Et Et {Cx,t+1 ux (Ct+1 )} Mt+1 1 ( ) = 1 Et Et {Cx,t+1 ux (Ct+1 )} Mt+1 ( ) 1 ! Et M 1 ( t+1 ) . = 1 Et Mt+1
( )
Ft
(f) The result in part a and the cash-in-advance constraint imply that M = px Cx = P Cx C . Cx
From part e we therefore obtain the risk neutral forward exchange rate, eq. (107) in Chapter 8, by again using the assumption that outputs and monies are independently distributed random variables: Mt+1 1 Et Mt+1 Mt+1 ( ) 1 Et Mt+1 ( ) ( ) 1 Ct+1 Et Cx,t+1 Et Et+1 Mt+1 Cx,t+1 ( ) ( ) 1 Ct+1 Et Et Cx,t+1 Mt+1 Cx,t+1 ( ! ) Cx,t+1 Ct+1 Et Et+1 Mt+1 Cx,t+1 ( ! ) Cx,t+1 Ct+1 Et Mt+1 Cx,t+1 Et {Et+1 /Pt+1 } . Et {1/Pt+1 } 92
(
Ft =
Why is there no risk premium? Here, the marginal utility of consumption is conditionally uncorrelated with the exchange rate when output and money shocks are independent. But if that correlation is zero, there is no foreign exchange risk premium. The last two results asked for in this part of the exercise follow from parts c and d above. 7. (a) The two-week average exchange rate series, sampled biweekly, does not follow a random walk even though the weekly exchange rate does. This can be shown very easily:
e Et Et+2 n o
= Et = Et = Et
n n n
1 (E 2 t+2 1 (E 2 t+1 1 (2Et 2
+ Et+1 )
+ t+2 + Et+1 )
+ 2t+1 + t+2 )
(b) In this case the problem does not arise. As long as the weekly exchange rate follows a random walk, a series formed by taking point-sample observations every other week also follows a random walk. 8. (a) The weekly series of two-week prediction errors is serially correlated. Let us suppose that the spot rate follows a random walk, et = et1 +t , where t is a serially uncorrelated white noise disturbance term. Then Cov {et+2 ft,2 , et+3 ft+1,2 } = E {(et+2 et )(et+3 et+1 )} . The preceding equality follows from the denition of covariance and the two assumptions that ft,2 = Et et+2 and that the spot exchange rate follows a random walk. Since et+2 et = t+2 + t+1 and et+3 et+1 = t+3 + t+2 , the preceding equality can be written as Cov {et+2 ft,2 , et+3 ft+1,2 } = E {(t+2 + t+1 )(t+3 + t+2 )}
2 = E t+2 6= 0.
e = Et 6= Et 1 (Et + Et1 ). 2
93
The reason for the nonzero covariance is that temporally adjacent overlapping multiperiod forward-rate errors share common innovations. They share innovations because the maturity of the forward contract (two weeks) is longer than the sampling interval (one week). We can show through similar steps that for j > 1, Cov {et+2 ft,2 , et+j+2 ft+j,2 } = 0. (b) A series sampled biweekly would not be serially correlated. This result follows easily from the argument of part a. In this case there are no overlapping multiperiod forecast errors. (c) We outline how a General Method of Moments (GMM) estimator could be used to test the hypothesis that Et (et+2 ft,2 ) = 0. Dene the two-periodahead forward-rate forecast error as t+2,2 et+2 ft,2 . The null hypothesis states that the forward rate is equal to the conditional expectation of the two-period-ahead spot rate, and is therefore the best (most ecient) unbiased predictor of the spot rate. This property implies that t+2,2 will be uncorrelated with any information available at time t. One possible way to test this implication is to run the following regression [which is similar to equation (110) in the text], et+2 et = a0 + a1 (ft,2 et ) + t+2,2 , (1)
where the dierence ft,2 et is the forward premium on date t. The forwardmarket eciency test asks whether one can reject the joint null hypothesis that a0 = 0 and a1 = 1. Under the null hypothesis, t+2,2 equals the forwardrate forecast error and so the orthogonality conditions E{t+2,2 } = 0 and E{(ft,2 et ) t+2,2 } = 0 are satised for a0 = 0 and a1 = 1. The specication implies that the ecient GMM estimator of a0 and a1 is the same as the ordinary least squares (OLS) estimator based on all the available (weekly) observations of spot and two-week-ahead forward rates. However the usual OLS standard errors would not be appropriate. When we use the weekly series of two-week-ahead prediction errors, we face a problem of serial correlation in t+2,2 (as discussed in part a). 94
To describe the GMM estimator, dene the coecient column vector = [a0 a1 ]0 , along with the vectors xt
1 ft,2 et
and
By eq. (1) above,
wt () xt t+2,2 () =
t+2,2 () (ft,2 et ) t+2,2 ()
t+2,2 () et+2 et [a0 + a1 (ft,2 et )] . The ecient GMM estimator (EGMM) for a sample of size T is derived as EGMM () = argmin w()0 1 w(), where w()=
T 1X wt (), T t=1 X
j ,
j=
and is a consistent estimate of . Under standard regularity conditions, the estimate EGMM is asymptotically normal with mean . One can estimate the asymptotic covariance matrix of EGMM by T
T X
t=1
j = E {wt ()wtj ()0 } ,
xt x0t
!1
T X
t=1
xt x0t
!1
Under the null hypothesis the coecient vector is 0 = [0, 1]0 . One can perform a Wald/Likelihood Ratio/Lagrange Multiplier test of the joint null hypothesis. (For details on GMM estimation and hypothesis testing, see R. Davidson and J. G. Mackinnon, Estimation and Inference in Econometrics, Oxford, 1993.) 95
Workbook2
Chapter 9 Solutions
1. (a) In the problem, the government unexpectedly freezes the money supply, which had previously been growing predictably at the proportional rate , at the previous periods level m, say. Since the economy was initially in a steady state, we have that p0 = m + (1 + ), which, together with eq. (12) in Chapter 9, implies that q0 = e0 m (1 + ). (1)
(Recall that y, i , p 0. This problem diers from the one in section 9.2.5 in that here, there is a forecast error in the level of the date 0 money supply, whereas in the chapter, there is no surprise concerning the date 0 money supply.) Let us reproduce eq. (18) from Chapter 9 (which assumes a constant long-run real exchange rate): et ef lex = t 1 (qt q). 1 +
For t = 0, we may use eq. (1) to eliminate q0 from the preceding equation: e0 ef lex = 0
1
1 [e0 m (1 + ) q] . 1 +
96
Observe that ef lex , the post-disturbance exible-price exchange rate, simply 0 equals q + m (because the money supply has been frozen at its date t = 1 level). The preceding equation therefore becomes 1 1 f lex (1 + ) = e0 (1 + ). e0 = q + m + + The eect on the real exchange rate (relative to the economys initial path) is the dierence e0 p0 q = e0 [m + (1 + )] q, which, from the last displayed equation equals 1 1 + (1 + ) (1 + ) = (1 + ). + + Thus, there is an initial real appreciation on date 0, after which the real exchange rate converges back to its long-run level q according to eq. (13) in Chapter 9. [The real appreciation is proportional to the total monetary shock, equal to the money-supply level shock, , plus the money-supply growth rate eect, , with the same proportionality factor as in eq. (17), Chapter 9.] From the aggregate demand equation in the chapter, eq. (3), we see that output falls initially by a percentage equal to times the percentage real appreciation, and then gradually rises back to its full-employment level. Footnote 17 in the chapter implies that the initial Home-less-Foreign real interest dierential is the expected change in q after the shock hits. Equation (13) in the chapter shows that q1 q0 = q1 q (q0 q) = (q0 q) 1 + = (1 + ). + Thus Homes relative real interest rate rises, a result that also can be derived from eq. (24) in Chapter 9. (b) To avoid the short-run disruptive eects, the monetary authority could unexpectedly increase the money supply by (1 + ) percent on date 0, while 97
reducing money growth between periods 1 and 0, 2 and 1, etc., to zero. The ination rates p1 p0 , p2 p1 , etc., would still all be zero, as in the strategy described in part a; p0 p1 would still equal as in part a. With the date 0 money supply equal to m+(1+) rather than m, however, p0 = m+(1+) would equal the new long-run price level. The diculty with such a plan is its credibilityhow does one convince markets in period 0 that the inationstabilization plan is serious without slowing money-supply growth in period 0 itself? Achieving such credibility is critical, for if the market does not believe future money growth will be curbed, the expected ination term in the Phillips curve will remain active and the liquidity squeeze and recession will occur in period 1 (when the money supply nally is frozen) rather than in period 0. Interestingly, an equivalent way to proceed would be simply to suddenly peg the exchange rate at the level q+ p0 = q + m + (1 + ), while simultaneously announcing that the exchange rate is henceforth irrevocably xed. If credible, such an announcement will bring the Home nominal interest rate down to the Foreign level as capital inows swell the economys money supply and prevent a liquidity squeeze. The credibility problems raised by this exchange-rate based ination stabilization approach are the subject of a large literature (since the approach has been tried so often, frequently with unfortunate nal results). For a survey see Carlos A. Vgh, Stopping e high ination: An analytical overview, International Monetary Fund Sta Papers 39 (September 1992): 626-95. 2. To do this question, we shall rst nd the generalized saddle path relation between the nominal and real exchange rates that holds when the long-run real exchange rate can vary exogenously (for example, due to shifts in foreign demand for domestic exports). Let us begin by reformulating eq. (3) in Chapter 9 as d yt = y + (et + p pt qt ), (2) so that we allow for a time-varying q. Following the logic that led to eq. (6)
98
in the chapter, we now obtain instead:

d pt+1 pt = (yt y) + et+1 et (qt+1 qt ). d Using eq. (2) above to eliminate yt y here, and using the denition of the real exchange rate, we nd that the real exchange rate follows the dierence equation qt+1 qt = (qt qt ) + qt+1 qt .
Since this equation can be expressed entirely in terms of the deviations q q, its solution is the same as eq. (13) in the chapter, but with a variable long-run real exchange rate: qs qs = (1 )st (qt qt ). (3)
Following the steps leading to eq. (9) in the chapter and assuming that the money supply is constant at m yields et (1 )qt qt + m . et+1 et = Dene et as the exible-price equilibrium exchange rate. (This variable is the same as the exible-price exchange rate as dened on p. 618, ef lex but t it will prove easier on the eyes to use the overbar notation instead.) Since monetary equilibrium condition (2) from the chapter also holds in a hypothetical exible price equilibrium, it is also true that the last equation holds with overbars, et (1 )qt qt + m . (4) et+1 et = Now subtract this equation from the one preceding it to obtain the analog of the equation preceding eq. (14) in Chapter 9: et et = 1 (et+1 et+1 ) + (qt qt ) . 1+ 1+ 99
Now we can calculate the economys saddle path as in the book, utilizing eq. (3) from above rather than eq. (13) in Chapter 9. The resulting generalized saddle path relation [after solving forward, excluding bubbles, and using (3) above] is 1 (qt qt ) , et et = (5) 1 + which generalizes eq. (16) in Chapter 9. We can nally proceed to answer the original question concerning an anticipated future rise in the equilibrium real exchange rate. Solve equation (4) forward (assuming no bubbles); the result is st 1 X et = m + qs . 1 + s=t 1 +
The exercise assumes that the economy starts out in a steady-state equilibrium on date t = 0, when it is announced that the equilibrium real exchange rate will rise from q to q0 on date T in the future. Thus, the preceding equation implies that the ex-price nominal exchange rate jumps up on date t = 0 from m + q to " T # T e0 = m + 1 q+ q0 > m + q. 1+ 1+ Using eq. (5) above and the fact that p0 = m, we derive 1 + [e0 (m + q)] = q0 q > 0. e0 (m + q) = + We have now assembled enough information to characterize the economys path after the initial real and nominal currency depreciations on date 0 (which induce higher aggregate demand and a rising price level). The economys entire path is illustrated in gures 9.1 and 9.2, where the over- and undershooting cases are shown separately. According to eq. (3) above, the domestic currency appreciates in real terms between dates 0 and T as the initial gain in competitiveness erodes. Notwithstanding this eect, eq. (5) 100
above implies that the currency continues to depreciate in nominal terms between dates 0 and T . During that time interval, according to eq. (5) above, " T t # T t 1 et = m + 1 (1 )t (q0 q) , q+ q0 + 1+ 1+ 1 + an expression that rises as t T . (We know the domestic currency must be depreciating at t = 0 because the impact rise in output, given the predetermined price level, implies a higher domestic nominal interest rate and therefore an expected domestic currency depreciation. However, if the preceding displayed equation satises det /dt > 0 when evaluated at t = 0, it must also satisfy that inequality for 0 < t < T , because, as you can easily see, d2 et /dt2 > 0 until t = T. Thus, the currencys depreciation actually accelerates prior to date T .) What happens on date T itself? The anticipated rise from q to q0 occurs, the price level jumps downward (because the exogenous change in demand was fully anticipated), and, according to eq. (3) above and the unnumbered equation preceding it, the real exchange rate jumps upward by the amount qT qT 1 = (q0 q) (1 )T 1 (q0 q) > 0. By eqs. (3) and (5), the nominal exchange rates movement is 1 1 0 (q q) (1 )T 1 (q0 q). eT eT 1 = 1+ 1 + Starting on date T, eq. (5) gives the nominal exchange rates path as et (m + q0 ) = According to eq. (3) above, qT q0 = (1 )T (q0 q) > 0. 101 1 (1 )tT (qT q0 ) . 1 + (6)
Equation (6) above therefore implies that for 1 > 0 (the overshooting case), the date T nominal exchange rate is above its new long-run level m + q0 , but that the gap is reduced over time (implying nominal currency appreciation from date T on). For 1 < 0 (the undershooting case) eT stands below its new long-run level m + q0 , but subsequently the currency depreciates in nominal terms toward its steady-state value. In both the overand undershooting cases, the real exchange rate is above q0 on date T , and therefore the currency appreciates in real terms afterward (that is, q falls after date T, asymptotically approaching q0 ). 3. (a) Since the only factors bueting the economy are the exogenous shocks and , since these are mean-zero i.i.d. shocks, and since the model contains no intrinsic persistence mechanisms, the economy is always expected next period to be at the equilibrium associated with the realizations = = pt Et1 pt = 0. It is easy to show that this equilibrium is characterized by e = p = m and y = i = 0. To solve for the equilibrium more generally, note that because we therefore have that Et1 pt = m, the equality ys = yd can be written (pt m) = (et pt ) + t . Since Et et+1 = m, the interest parity condition is it+1 = m et ; substituting this into the monetary equilibrium condition gives (1 + + )m (1 + )pt = et + t . Combining the preceding two equations to solve for e and p, and then using the aggregate supply schedule to calculate equilibrium output y, we nd that et = m pt = m + (1 + )t + ( + )t , (1 + ) + ( + )
t t , (1 + ) + ( + ) (t t ) yt = . (1 + ) + ( + ) 102
Using the last of these solutions, we see that Et1 {yt }2 which equals the variance of output, since Et1 yt = 0is given by Et1 {yt }2 =
2 2 2 (2 + 2 ) . [(1 + ) + ( + )]2
(Recall the assumption that and are distributed independently of each other, making their covariance zero.) (b) When the exchange rate is xed at e, Et1 pt = e = m. The equality ys = yd therefore can be written as (pt e) = (e pt ) + t , implying that pt = e +
2 t t 2 . , yt = , Et1 {yt }2 = + + ( + )2
2 2 (c) The limiting behavior /v 0 means that the variance of real (aggregatedemand) shocks is becoming negligible relative to that of monetary shocks. But the xed exchange rate policy in part b eliminates the eect of the monetary shock on output! Any changes in money demand are automatically accommodated through foreign-exchange intervention, with no side-eects on output. A xed exchange rate thus is bound to be better when moneymarket shocks dominate. More formally, combining the bottom line results in parts a and b, we see that the ratio
2 2 2 2 As /v 0, / , and thus the variability of output under a constant money supply rule becomes arbitrarily large compared to that under a constant exchange rate rule.
2 2 2 Et1 {yt }2 |flex ( + )2 2 = / . 2 + Et1 {yt }2 |fix [(1 + ) + ( + )]
103
(d) [There are two typos in the statement of this part of the exercise. The monetary policy rule should be as given below, rather than mt m = (et e) as in the book. Also, the optimal value of lies between and , not 0 and .] To analyze a monetary policy rule of the form mt m = (e et ), we use the following trick. We have already noted that in this particular model (with i.i.d. shocks), we may write the interest parity condition as it+1 = e et , so that the money-market equilibrium condition is mt pt = (e et ) + yt + t . However, if we now substitute the policy rule for mt into this money-market equilibrium equation, we can write the latter as m pt = ( + )(e et ) + yt + t . In short, the intervention policy rule leaves us with a model isomorphic to a model with a xed money supply in which the interest semi-elasticity of money demand is + . An implication is that we can immediately use the result of part a to write the variance of output under the intervention policy as Et1 {yt }2 =
2 2 2 (2 + 2 ) . [(1 + ) + ( + )]2
To nd the optimal value of the synthetic parameter , we compute the derivative

2 dEt1 {yt }2 [(1 + ) + ( + )] (2 + 2 ) ( + ) 2 2 = 22 d [(1 + ) + ( + )]3
104
and set it equal to zero. (We leave it to the student to check that the relevant second-order conditions are satised.) The resulting condition simplies considerably, to 2 (1 + ) ( + ) 2 = 0, 2 which is readily solved to yield:
2 ( + ) = . 2 (1 + )
Finally, using the denition of above, we nd that the optimal monetary reaction parameter is 2 ( + ) = . 2 (1 + ) It is instructive to consider how this optimal rule embodies the answer to part c above. Notice that is monotonically increasing in the variance ratio 2 2 / . The greater the variance of monetary compared with real shocks, the greater the tendency for intervention policy to lean against the wind. As 2 2 / , as well, meaning that we have a xed exchange rate: any rise in e over e, for example, elicits a large monetary contraction that drives e immediately back to its target level. Leaning against the wind is 2 2 not always an optimal response, however. As / 0 (in which case real shocks are preponderant), according to the preceding formula. It becomes optimal for intervention policy to accentuate rather than resist the exchange rates movement. The solutions for the exchange rate, price of domestic output, and output levels (recall part a, above) show why this is so. A policy that sets = eectively sets the interest semi-elasticity of money demand equal to zero. The solutions in part a show that in that case, the shock aects the exchange rate, but not output or domestic prices. The role of monetary policy is to ensure that, when world demand for domestic goods rises (say), the currency appreciates by enough to forestall any increase in output. That policy reduces the variance of output to zero. 105
4. (a) Proceeding as in section 9.5.2.2 in the book one nds that

e t =
k k Et1 {t } =
and
e t = t
[t Et1 {t }] (t 1) zt zt e + = t + . 1+ 1+ 1+ 1+
Mean ination is the same as when is known to equal 1, but the ex post uncertainty in policymaker type creates extra ination variability. (b) To calculate expected social loss, assume Cov(, z) = 0 and compute ( 2 (t 1) zt Et1 Lt = Et1 + zt k 1+ 1+ 2 ) zt k (t 1) + + 1+ 1+ = k2 +
2 2 2 2 2 2 z (k )2 z 2 + + + + (1 + )2 (1 + )2 (1 + )2 (1 + )2 2 2 2 2 z (k ) + + . = k2 + 1+ 1+
2 (c) When = 0, the clear solution is = k (the Walsh 1995 solution); but with unpredictable, there is a trade-o between reducing mean ination by choosing a positive and raising the variance of ination because the policymakers preferences are random. From the rst order condition for minimizing the expected loss in part b with respect to we can solve for as: (1 + )k = . 2 1 + (1 + )
Thus, greater uncertainty about the weight reduces the optimal below the benchmark value of k.
106
5. (a) (Note: In this problem the reader can simplify by setting k = 0, without substantively changing the nature of the exercise. However, the e solution here is for an arbitrary k.) Taking t as given, dierentiate Lt with respect to t to solve for the one-shot-game equilibrium level of t . The rst-order condition is t + t = 0, implying that t = t in equilibrium. With rational expectations,
e t = Et1 t = Et1 t = 1 2
0 + 1 2 = 1. 2
In part b we will need to know the expected loss in the discretionary onee shot-game equilibrium. Because t = 1,we can calculate it as Et1 Ld = Et1 t (t 1 k) + 1 2 t t 2 Et1 Ld = 2 + 1 + k + 1 2 = k. t 2 (b) The reaction function must satisfy
1 2
Since Et1 2 = t to
1 2
0 + 1 4 = 2, however, the preceding expression simplies 2
(0) + 1 (2) = 0 2
and minimize Et1 Lc = t

1 2
Substituting the constraint of zero expected ination into the loss function, we write the problem to be solved as the unconstrained minimization of Et1 Lc = t
1 2
(0)2 + 1 2 [(2) k] + 1 (2)2 . 2 2 2
(2)2 + 1 2 [(2) k] + 1 (2)2 . 2 2 2 107
The rst-order condition for a maximum is dEt1 Lc t = 1 (2) 1 + 1 (2) = 0, 2 2 d(2) implying that (0) = 1 and (2) = 1. Given this policy rule, Et1 Lc = t =
1 2 1 4
1
2
+ (k 1) +
1 2
(1)2 + 1 2 (1 k) + 1 (1)2 2 2 <

1 4 Et1 Ld t
= k
= k.
(c) Suppose rst that the central bank always reveals t on date t 1 before e e t is set, so that t = t . Its expected loss (from the standpoint of date t 2, when it itself doesnt yet know its value of t ) is Et2 Lyes = t
1 2
0 + 1 (2k + 2) = k + 1. 2
e If it does not reveal t then t = 1 and the expected loss is
Et2 Lno = t
1 2
0 + 1 [2(1 k) + 2] = k. 2
Since Et2 Lyes > Et2 Lno , the central bank would prefer to precommit itself t t e not to reveal t before the public sets t . That is, the central bank prefers a system that mandates central-bank secrecy about its true preferenceswhat has sometimes been characterized as monetary mystique. (This result is heavily dependent on the specic form of loss function assumed here.)
108
Workbook2
Chapter 10 Solutions
1. (a) With a positive steady-state gross money supply growth rate of 1 + , eq. (26) in Chapter 10 is replaced by m0 = m = 0 1 1 (1 + ) (1 + ) y0, (260 )
where (for this problem only) m M/P denotes the real money supply. The value for y 0 is the same as in the case with zero steady-state money growth. The log-linear version of the money demand equation becomes mreal,t = ct 1 pt+1 , (370 ) rt+1 2 (1 + ) (1 + ) 1 (1 + ) (1 + ) (1 + )
where mreal,t = dmt /m denotes the percentage deviation of real money balances from their steady-state level, pt [(Pt /Pt1 ) /(1+)]1 denotes the percentage deviation of ination from its (gross) steady-state level of 1 + , and the other variables are as dened in the text. The foreign counterpart to (370 ) is
m real,t = ct
1
1 p . (380 ) rt+1 2 (1 + ) (1 + ) 1 t+1 (1 + ) (1 + ) (1 + )
109
Denote by t (a boldface ) the percentage deviation of nominal money growth from its steady-state value. Note that t = mreal,t + pt . Also, by consumption-based purchasing power parity, pt = p + et , t where et is the percentage deviation of the growth rate of the nominal exchange rate, Et /Et1 , from its initial date-zero steady-state value of unity (recall that = ). It is then possible to derive t et = ct c ct1 c t t t1

1 (et+1 et ) . (1 + ) (1 + ) 1 (390 )
Solutions for steady-state equilibrium ination and nominal exchange rate growth follow immediately from eqs. (370 )-(390 ): p = 0, p = 0, = e.
(500 ) (510 ) (520 )
(b) Assume a permanent unanticipated rise in the home rate of money growth occurring on date 1, with prices preset a period in advance and adjusting to their exible-price level after one period, absent new shocks. Given that in the initial steady state the exchange rate is expected to remain constant (because initially, = ), it follows that e = e, where e is the percentage deviation of the nominal exchange rate from its preshock steady state leveli.e., its level along the economys steady-state 110
path. (As in the text, sans serif variables with overbars denote new postshock steady-state values, for period 2 and beyond. Variables without bars denote postshock date 1 values; thus e e1 e0 .) Note that e can also be interpreted as the percentage by which the nominal exchange rate would change on impact if prices were fully exible, so that e = ef lex , (That is, ef lex is the percentage deviation of E f lex from its pre-shock steady state level, with E f lex being the nominal exchange that would obtain on impact if output prices were fully exible. ) It is then possible to rewrite eq. (390 ) as: (1 + ) (1 + ) 1 (c c ) , (600 ) e = ef lex (1 + ) (1 + ) where ef lex = , with = 0 and c c = c c . Figure 10.1 shows eq. (600 ) as the downward sloping MM schedule. Notice that the short-run nominal exchange rate, e, is going to be less than the value that would obtain if prices were fully exible, ef lex , because the rise in the Home rate of money growth relative to Foreigns entails a short-run increase in the consumption growth dierential. As in the text, it is possible to derive a second schedule in e and c c using the short-run equilibrium conditions other than the money demand equations, together with eq. (45) of Chapter 10, and recalling that on impact e = e. Denote by Pt the percentage deviation of pt (h)/Pt from its preshock steady-state level of unity, and by pt (h) the percentage deviation of pt (h)/pt1 (h) from its pre-shock steady-state level of 1 + . It is easy to show that p = (1 n) e, and P P = e. Following the same steps as in the chapter, one then obtains e= (1 + ) 2 (c c ) , (2 1) (640 )
which is the upward-sloping schedule GG in gure 10.1. The GG locus has a positive slope because Homes consumption growth can rise relative 111
to Foreigns in the short-run only if the growth in the nominal exchange rate increases, allowing Homes output to rise relative to Foreigns. The intersection of the two schedules gives the equilibrium nominal exchange rate at the time of the shock. Note that the level of c c given by the diagram is permanent, but eq. (520 ) must be used to calculate nominal exchange rate growth after the initial, sticky-price period. 2. The nominal home-currency price of a nontraded good is Pn , and that of a traded good is Pt . Therefore, the real price of nontraded in terms of traded goods is Pn /Pt , and the consumption-based price index corresponding to the utility function specied in the problem is 1 / (1 )1 (expressed in units of traded goods). Since Pn is xed in the short run, and since an unanticipated money-supply increase depreciates the domestic currency making Pt rise, it simultaneously lowers 1 . The resulting change in the log of the consumption-based price index (measured in tradables) is approximately (1 )Pt . Notice that this change corresponds (albeit with the opposite sign) to the change in the countrys real exchange rate, q = EP /P , where P is the home consumer price index and P is the rest-of-world consumer price index measured in foreign currency units. Since it is assumed that purchasing power parity holds for traded goods (Pt = EPt ),
EP /EPt P /Pt EP = = . q= P P/Pt P/Pt
Because P/Pt =
1 (1 )1
(given the assumed utility function) and because P /Pt is not aected by the shock in the small country, the absolute change in 1 equals the increase in qwhich is a real depreciation of the domestic currency.
112
The consumption-based real interest rate is given by 1+

c rt+1
(1 + r) (Pn,t /Pt,t )1 = . (Pn,t+1 /Pt,t+1 )1
(Refer to section 4.4.1.3 in the book to see how this expression is derived. The world interest rate r is the own-rate on tradables in section 10.2.) Money is completely neutral in the long run in this particular model (but only because the special structure of the model implies that an unanticipated money shock has no current-account eects.) Therefore, Pn,t+1 /Pt,t+1 is unaected by the money shock and the direction in which the consumption-based real interest rate moves in the short run is the same as the direction in which Pn,t /Pt,t moves. So, in fact, an unanticipated domestic money-supply increase causes a fall in the home real interest rate and a real depreciation of the home currency, just as in the Dornbusch model. It is easy to see that this result extends to money growth shocks, since it is still the case that any real eect on Pn,t /Pt,t dies out after just one period. 3. Let py be the home-currency price of the single good exported to the rest of the world, P the home-currency price of the imported good. Assume that P = EP , where P is the rest-of-world price index and E is the home-currency price of the rest-of-world currency. One can then rewrite the demand curve faced by the small country as: yd =
py P
C w.
The utility function of the small countrys representative agent is Ut =

X st log C s=t s + log
Ms 2 ys , Ps 2
(10 )
where C is consumption of the single imported good. The period budget constraint is Pt Bt+1 + Mt = (1 + r) Pt Bt + Mt1 + py,t yt Pt Ct Pt t , 113 (80 )
where r is the constant world net interest rate, with (1 + r) = 1. Throughout, a change in the foreign-currency price of the good produced by the small country is assumed to have, ceteris paribus, a negligible eect on the foreign-currency world price index P . The rst-order conditions for the maximization problem of the small-country representative agent are Ct+1 = Ct , Mt 1 + it+1 = Ct , Pt it+1
+1
(130 )
!
(140 ) (150 )
yt =
1 1 w 1 , (Ct ) Ct
where 1 + it+1 = (1 + r) Pt+1 /Pt and the usual transversality condition must hold. Assuming an initial symmetric steady-state where py = E P and C = w C , so that py = P and B = 0, and log-linearizing around that steady-state, one obtains the following log-linear versions of the rst-order conditions, ct+1 = ct , mt et = ct 1 (et+1 et ) , r (350 ) (370 ) (330 )
( + 1) yt = ct ,
where it is assumed that cw = p = 0 on every date, so that pt = et , and yt = (et py,t ). The log-linearized version of the economy-wide resource constraint is then bt+1 = (1 + r) bt + ( 1) (et py,t ) ct . Following the same steps as in the text, one can show straightforwardly that in the steady-state, 1+ c= (450 ) rb, 2 1 py,t et = rb, (460 ) 2 114
m e = c.
(520 )
If one assumes that the small-country-currency price py of the export good is set one period in advance, and reverts to its exible-price level after a single period absent new shocks, then, given that c = c and m = m, it follows from eq. (370 ) that e = e. Moreover, since in the short run y = e, and thus b = ( 1) e c, one derives the following schedule in e and c, r (1 + ) + 2 e= c, (640 ) 2 1) r ( which, together with the schedule e = m c, gives the equilibrium exchange rate r (1 + ) + 2 e= m. (650 ) r (1 + ) + 2 Because = (1 )/ = r, this last equation is the same as eq. (65) in the text when the level of the money supply in the rest of the world is constant. 4. [In the cash-in-advance constraint given in the statement of this exercise, j Cn,t (z) should be cj (z).] Individual js period budget constraint is: n,t
j j j Pt,t Bt+1 + Mtj = Pt,t (1 + r)Bt + Mt1 + pn,t (j)y n,t (j) j j + Pt,t y t Pn,t Cn,t Pt,t Ct,t Pt,t t .
(1)
A nontraded-goods producer j faces the demand curve pn (j) a = Cn . (2) Pn The rst-order conditions are found by maximizing the lifetime utility function subject to (1), (2) and the contemporaneous cash-in-advance constraint: y d (j) n Bt+1 : Cn,t : Ct,t = Ct,t+1 , Cn,t =
+1
"
Pt,t Ct,t , Pn,t

1
y n,t :
yn,t =
( 1) Ct,t
a Pn,t Cn,t
Pt,t
115
Substitute for Ct,t in the rst-order condition for yn,t using the rst-order condition for Cn,t to obtain y n,t
+1
a ( 1)(1 ) Cn,t = . Cn,t
"
(3)
Once consumers have determined the amount of traded and nontraded goods they wish to consume, and the price at which they will sell their output, the cash-in-advance constraint determines the amount of money they wish to hold. This is because money does not enter the utility function. As long as the nominal interest rate on bonds is positive, people will never want to hold any money in excess of what they require to nance current consumption. (a) Flexible price case: In the symmetric market equilibrium, Cn,t = yn,t = a Cn,t for every nontradable good z. Thus equation (3) implies that in the exible-price equilibrium, ( 1)(1 ) yn =
" #1
2
(4)
(b) The monopoly level of output of each nontraded good is too low. As discussed in the text on p. 668, a planner equates the marginal utility of composite nontradables consumption with the marginal welfare cost of higher output in terms of forgone leisure. Assuming the planner gives all agents equal weight, his problem can be written as that of maximizing Ut =
X s=t
st log yt + (1 ) log yn,s
(yn,s )2 . 2
(The planner internalizes the constraints that in a symmetric allocation, a yn,t = Cn,t = Cn,t and yt = Ct,t .) The rst-order condition with respect to yn,s gives the optimal level of nontraded goods output that the planner will choose, (1 ) = yn , yn 116
which yields yn plan
(1 ) =
"
#1
2
(5)
The planner will therefore choose the level of output at which the marginal utility from consumption of nontradables is equal to the marginal cost the leisure forgone in producing them. Plainly this output level exceeds that in part a, see eq. (4). (c) The cash-in-advance constraint holds with equality in equilibrium: Mt = Pt,t yt + Pn,t yn,t . Furthermore, Ct,t = yt . (7) In the exible-price case, the money price of nontraded goods is found by combining the rst-order condition for Cn,t with eqs. (6) and (7): Pn,t = (1 )Mt . yn,t (8) (6)
Given the symmetry of the model, the period t money price of every nontraded good will be set at the level pn,t at which, in the absence of monetary surprises, each producers output would be given by yn in eq. (4). Thus, (1 )Mte . pn,t = Pn,t = yn (9)
Given temporarily xed nontraded goods prices, short-run output is demand determined, that is, yn,t = Cn,t . Using this result, together with (8) and (9), we obtain the solution for yn,t : yn,t = Mt yn Mte (10)
(d) Period t monetary policy aects only period t welfare in the one-shot game. Also, Ct,t = yt in equilibrium. The monetary authorities therefore 117
set Mt to maximize the expression specied in the problem since the other elements of the representative agents objective function are exogenous. (e) Eliminate Cn,t from the monetary authorities objective function using eq. (10) above. Observe also that the price of tradables always moves proportionally to the money supply, ceteris paribus, because the rst-order condition for Cn,t and eq. (6) together imply that Pt,t = Mt . yt
!2
One therefore can express the authorities maximization problem as: max (1 ) (log Mt + log yn log Mte )
Mt
Mt yn Mte
!2
Mt Mt1
Take the derivative with respect to Mt to get the solution for the one-shotgame equilibrium level of money growth: yn (1 ) Mt Mte
!2
Mt Mt = 0. 2 Mt1
Now impose the usual condition for a time consistent equilibrium, Mte = Mt , which implies that yn,t = yn . Making use of eq. (4) for steady-state output yn , we obtain the solution for equilibrium ination Pt,t Pn,t Mt = = = Mt1 Pt,t1 Pn,t1
!1
2
(11)
But the right-hand side of eq. (11) above can be written in the alternative 1 form { [(n )2 (n )2 ] /} 2 ; to see why, simply substitute for yn and yn y plan y plan using eqs. (4) and (5) from parts a and b.
118
Figure 6.1
Payment, P()
-e e Shock,
Figure 6.2
Period 2 consumption, C2
GDP
GNP D
U ( = 0)
UD
{
ID
Period 1 consumption, C 1
Figure 6.3
GDP
GNPN A' GNP D A UN > UD
{
ID
Figure 6.4
R*
GDP
GNP R B GNP D A
UN = UD
{
IR
Figure 7.1
Consumption per capita, c c = 0
A B
S'
c NEW
S
A'
k = 0 kNEW = 0
-g/(1+n)
kNEW
Capital-labor ratio, k
Figure 9.1
Log nominal exchange rate, e qNEW = 0 eNEW = 0 SNEW SNEW e
q'
Log real exchange rate, q
Figure 9.2
Log nominal exchange rate, e qNEW = 0 SNEW
SNEW
eNEW = 0
q'
Log real exchange rate, q
Figure 10.1
Percent change in exchange rate, e M' - * = e G

flex
M M' Percent change in relative domestic consumption, c - c* G M

Or Solutions

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Or Solutions

Uploaded by

Copyright:

Available Formats

FOUNDA TIONS OF INTERNA TIONAL

c MIT Press, 1998

Foundations of International Macroeconomics1

U(C1 , C2 ) U (C1 , C2 ) b b r(Y1 C1 ) = r (Y1 C1 ) . C2 C1

U(C1 , C2 ) U(C1 , C2 ) b (Y1 C1 )dr = r (Y1 C1 ) . C2 C1

2. (a) Substitution of the consumption Euler equation C2 = (1 + r)C1 2

1 +r > 0. 1 + r(1 + r)2

or C2 = (1 + r) C1 . As 0, the Euler equation approaches C2 = C1 . 5. The global equilibrium condition is

C1 (R, W1 ) dW1 1 dR W1 dY1 # =" dY1 dC1 (R, W1 ) dC1 (R, W1 ) + dR dR

1 A F (I1 ) + I1 Y1 I1 + 2 I1 , 1 + (1 + r)1 1+r

The eect on Foreigns current account CA = Y1 C1 I1 is 1

where dr is as in eq. (5) above except for the obvious changes:

From this equation,

h (C2 )C2 C1 = = . R 1+R 1+R

u [C1 , Y2 + (1 + r)(Y1 C1 )] subject to the goods-market equilibrium condition

Foundations of International Macroeconomics1

st 1 = r[1 + (1 )r]st Bt 1+r s=t rBt = = (1 + r)Bt . 1 + (1 )r 1 1+r

+2 (1 ) u(Ct ) + u(Ct+1 ) + 2 u(Ct+2 ) + ....

Using the intertemporal Euler equation, we thus obtain, (1 + r) = 1 = Et 1 Ct+1 Ct

Ct+1 1 = Et exp log Ct

1 Ct+1 1 Ct+1 = exp Et log + 2 Vart log Ct 2 Ct 15

1 Ct+1 Vart log 2 Ct

1 Vart {t+1 } + t+1 , 2

The current account identity gives

1 r X Bt+1 Bt = Yt + rBt Ct = Yt 1 + r s=t 1 + r

or, dividing by 1 + r and rearranging, as

1 1 L1 CAt = L1 Zt . 1+r 1+r

1 1 1 L1 CAt = (Et Zt+1 Zt ) = L1 Et Zt 1 1+r 1+r 1+r

lim (1 + r)T Bt+T +1 = 0

q1 AFKK K + ! < 0, = q1 1 r AFKK K +

Foundations of International Macroeconomics1

That of a middle-aged person is sm = (1 + e) y y

and that of an old person is

2 m y y 1 (1+g)yt2 3 3 y + g)2 yt2 + y m

Taking the derivative above with respect to g, we nd it is proportional to

concentrated on the middle-aged, Aggregate = saving rate =

which can be rewritten in per capita terms as kt+1 = wt sy t = . (1 + n) (1 + )(1 + n)

p,t1 p,t2 p Bt+1 = 1 bp,t + bt+1 + 2 bt+1 + ..., t+1

h p,t1 p,t2 1+r 1 bt + bt st (1 ) P

i (1 + r) h p,t1 p,t2 1 bp,t + bt + 2 bt + ... + Yt Tt Ct t i (1 + r) h p,t1 p,t2 1 bt = + bt + ... + Yt Tt Ct p = (1 + r)Bt + Yt Tt Ct .

p,t1 p,t2 1 bp,t + bt+1 + 2 bt+1 + ... t+1

and, for steady state aggregate consumption, C=

Substituting for taxes using formula (8), we nd that:

X 1 1+r (rn)dt + s = r s=t 1 + r

and the government constraint

u0 (C) Gs = 0, au0 (C)

X 1 r u0 (C) 1+r s=t 1 + r

Foundations of International Macroeconomics1

where k K/L. Log-dierentiation of eq. (1) yields, f

b Similarly we can solve for Ln as,

The current account identity is

(st) Rc t,s s=t

n,s c Rt,s Yt,s +ps YPs Is Gs

As in section 2.2.2 of the book, let us dene t

Then we can write the last equation as

Bt rc Bt Zt Bt+1 = t + Pt Pt1 Pt1 Pt rc Bt Zt = t + Pt1 Pt t 1 + t

c (1+rt )Bt Pt1

c (1+rt )Bt Pt1

Dene the permanent level of a variable as

c Zs s=t Rt,s Ps P c s=t Rt,s

c Zs s=t Rt,s Ps P c s=t Rt,s

e(r+)t A(T ) hdt =

Equating this to w/(r + ) yields the answer that was claimed. 42