Anda di halaman 1dari 109

Monetary Theory and Policy:

Problem Solutions
Carl E. Walsh
University of California, Santa Cruz
March 16, 1999

Contents
1 Chapter 2: Money in a General Equilibrium Framework

2 Chapter 3: Money and Transactions

3 Chapter 4: Money and Public Finance

20

4 Chapter 5: Money and Output in the Short Run

27

5 Chapter 6: Money and the Open Economy

43

6 Chapter 8: Discretionary Policy and Time Inconsistency

59

7 Chapter 9: Monetary-Policy Operating Procedures

81

8 Chapter 10: Interest Rates and Monetary Policy

97

9 Typos

109

Solutions 
c 1998 by Carl E. Walsh. Comments and corrections should be sent to walshc@cats.ucsc.edu.

Chapter 2: Money in a General Equilibrium


Framework
1. Calvo and Leiderman (1992): A commonly used specification of the demand for money, originally due to Cagen (1956), assumes m = Aeit
where A and are parameters and i is the nominal rate of interest. In
the Sidrauski (1967) model, assume that utility is separable in consumption and real money balances: u(ct , mt ) = w(ct ) + v(mt ), and further
assume that v(mt ) = mt (B D ln mt ) where B and D are positive parameters. Show that the demand for money is given by mt = Aet it
B
where A = e( D 1) and t = w (ct )/D.

The basic condition from which one can derive the demand for money in
Sidrauskis money-in-the-utility function model is given by equation (2.23) on
page 57. This equation states that the ratio of the marginal utility of money to
the marginal utility of consumption depends on the nominal rate of interest:
it
um (ct , mt )
=
it
uc (ct , mt )
1 + it
Notice that the expression has been simplified by employing the approximation
x/(1 + x) x for small x. Sidrauski developed his model in continuous time,
in which case the first order condition takes the exact form
um (ct , mt )
= it
uc (ct , mt )
Using the proposed utility function, um = B D D ln mt and uc = w (ct ), so
this condition becomes
B/D 1 ln mt
B D D ln mt
um
=
= it
=
uc
w (ct )
w (ct )/D
Rearranging yields
ln mt = (

w (ct )it
B
1)
D
D

or
mt = e( D 1) e
B

w (ct )
it
D


i
mt
2. Suppose u(ct , mt ) =
], > 0 and = 0.95. Asi=0 [ln ct + mt e
sume the production function is f (k) = k0.5 and = 0.02. What rate of
inflation maximizes steady-state welfare? How do real money balances at
the welfare maximizing inflation rate depend on ?

The steady-state welfare maximizing nominal rate of interest is iss = 0 (see


section 2.3.1.2, pages 61-64) at which point um = 0. If R is the gross real rate
of interest (one plus the real rate of interest), 1 + i = R(1 + ) and the rate of
inflation that yields a zero nominal rate of inflation is
ss =

1
1
R

In the steady-state, R is equal to 1/, or 1/R = = 0.95 (see equation 2.19,


page 54). Hence, the optimal rate of inflation is 0.95 1 = 0.5 or a 5% rate
of deflation.
To determine how money demand depends on the parameter , use the representative agents first order condition (see equation 2.23, page 57), evaluated
at the steady-state nominal rate of interest:
iss
um
=
uc
1 + iss
Given the form of the utility function, this becomes
iss
um
= ct (1 mt ) emt =
uc
1 + iss
At the welfare maximizing inflation rate, iss = 0, which requires 1 mss = 0
or
mss =

Thus, real money demand is decreasing in .


3. Assume that mt = Aeit where A and are constants. Calculate the
welfare cost of inflation in terms of A and , expressed as a percentage of
steady-state consumption (normalized to equal 1). Does the cost increase
or decrease with ? Explain why.
A traditional method for determining the welfare cost of inflation involves
calculating the area under the money demand function. That is, the loss in
consumer surplus when the interest rate is equal to i > 0 is given by
 i
Aex dx
l(i, A, )
0

Evaluating this integral yields


l(i, A, ) =


A
1 ei

(1)

as the welfare cost of an inflation rate of = i r if r is the real rate of return.


The effect of on this cost is
 

A(iei ) A(1 ei )
A  i
l(i, A, )
e (1 + i) 1 0
=
=
2
2

The sign depends on ei (1 + i) 1, but this is always negative ( ex (1 + x)


is maximized when x = 0 at which point ex (1 + x) = 1; it then declines with
x).
From the specification of the money demand equation, the interest elasticity
of money demand is i, so money demand is more sensitive to the interest
rate the larger is . As the nominal interest rate rises with an increase in inflation, households respond by reducing their demand for money, thereby helping
to reduce the distortion generated by the inflation tax. The greater the interest
sensitive of money demand, the lower will be the welfare cost of the inflation
tax.
4. Suppose a nominal interest rate of im is paid on money balances. These
payments are financed by a combination of lump-sum taxes and printing
money. Let a be the fraction financed by lump-sum taxes. The governments budget identity is t + vt = im mt , with t = aim mt and vt = mt .
Using Sidrauskis model,
(a) Show that the ratio of the marginal utility of money to the marginal
utility of consumption will equal r + im = i im . Explain why.
(b) Show how iim is affected by the method used to finance the interest
payments on money. Explain the economics behind your result.
(a) The budget constraint in the basic Sidrauski model must be modified to
take into account the interest payments on money and that net transfers ( in
equation (2.12), page 52) consists of two components, the first being the lumpsum transfer v and the second being the lump-sum tax . Thus, the budget
constraint becomes
1 + im
mt1 t + vt = ct + kt + mt
f(kt1 ) + (1 )kt1 +
1+
where population growth has been ignored for simplicity. The value function for
the problem is still given by (2.13), but the first order conditions change because
of the change in the budget constraint. In particular, equation (2.15) becomes
um [fk + 1 ] V ( t+1 ) +

(1 + im )
V ( t+1 ) = 0
1+

(2)

The first order condition for consumption (see 2.14), together with the envelope
condition (see 2.17) implies
uc (ct , mt ) = [fk + 1 ] V ( t+1 ) = [fk + 1 ] uc (ct+1 , mt+1 )
= Ruc (ct+1 , mt+1 )
4

Using this result, equation (2) can be rearranged, resulting in





(1 + im )
(1 + im ) uc (ct+1 , mt+1 )
um
1
=1
= 1
uc
1+
uc (ct , mt )
1+
R
R(1 + ) (1 + im )
i im
(1 + im )
=
=
= 1
R(1 + )
R(1 + )
1+i
The ratio of the marginal utility of money to consumption is set equal to the
opportunity cost of money. Since money now pays a nominal rate of interest
im , this opportunity cost is i im , the difference between the nominal return on
capital and the nominal return on money.
(b) From the governments budget constraint, interest payments not financed
through lump-sum taxes must be financed by printing more money. Hence, v =
m = (1 a)im m, or the rate of money growth will equal = (1 a)im . In
the steady-state, = . This means that = (1 a)im . Hence, the opportunity
cost of money is given by
i im r + im = r + (1 a)im im = r aim
where r = R 1. Paying interest on money affects the opportunity cost of
money only if a > 0. Printing money to finance interest payments on money
only results in inflation; this raises the nominal interest rate i, thereby offsetting
the effect of paying interest.


5. Assume u(c, m) = ca 1 + (m m)2 , a > 0. Normalize so that the
steady-state value of consumption is equal to 1 (css = 1). Using equation
(2.23) of the text, show that there exist two steady state equilibrium values
for real money balances if aI ss < 1. (Recall that I = i/(1 + i) where i is
the nominal rate of interest.)
From equation (2.23),
um (css , mss )
= I ss
uc (css , mss )
Using the utility function specified in the question,
um (c, m) = 2ca (m m)
and


uc (c, m) = ac(1+a) 1 + (m m)2
Therefore
2(css )a (m mss )
um (css , mss )
=
=I
ss
ss
ss
uc (c , m )
a(c )(1+a) [1 + (m mss )2 ]
5

(3)

We now need to solve this equation for mss . Let x (m mss ). Then equation
(3) becomes


2(css )a x = aI(css )(1+a) 1 + x2
If css is normalized to equal 1, this becomes


2x = aI 1 + x2
which can be rewritten more explicitly as a quadratic in x:
x2

2
x+1=0
aI

From the quadratic formula,


x=

2
aI

4
(aI)2

1
=

aI

1
1
(aI)2

There will be two real solutions if and only if


1
1>0
(aI)2
which holds for
aI < 1
If this condition is satisfied, both solutions for x are positive (so that mss < m ).
This can be verified by noting that aI < 1 implies

1
1

1>0
aI
(aI)2

6. In Sidrauskis money-in-the-utility-function model augmented to include


variable labor supply, money is superneutral if the representative agents
preferences are given by

i
d
i (ct+i mt+i )b lt+i
u(ct+i , mt+i , lt+i ) =
but not if they are given by


d
i u(ct+i , mt+i , lt+i ) =
i (ct+i + kmt+i )b lt+i
Discuss. (Assume output depends on capital and labor and the aggregate
production function is Cobb-Douglas.)

The steady-state values of css , kss , lss , yss must satisfy the following four
equations (see pages 65-67):
ul
= fl (kss , 1 lss )
uc

(4)

1
1+

(5)

css = f(kss , 1 lss ) kss

(6)

y ss = f(kss , 1 lss )

(7)

fk (kss , 1 lss ) =

d
If the single period utility function is of the form (ct+i mt+i )b lt+i
, then

dcb mb ld1
dc
ul
= b1 b d =
uc
bc m l
bl
is independent of m and equations (4) - (7) involve only the four unknowns css ,
kss , lss , y ss . These can be solved for css , kss , lss , and yss independently of m
or inflation. Superneutrality holds.
d
, then
If the utility function is (ct+i + kmt+i )b lt+i
d(c + km)b ld1
d(c + km)
ul
=
=
uc
b(c + km)b1 ld
al
which is not independent of m. Thus, equations (4) - (7) will involve 5 unknowns ( css , kss , lss , y ss , and mss ) and cannot be solved independently of the
money demand condition and inflation. Superneutrality does not hold.
7. Suppose the representative agent does not treat t as a lump sum transfer,
but instead assumes her transfer will be proportional to her own holdings
of money (since in equilibrium, is proportional to m). Solve for the
agents demand for money. What is the welfare cost of inflation?
This question is basically the same as Question 4. If the transfer is viewed by
the individual as proportional to her own money holdings, then this is equivalent
to the individual viewing money as paying a nominal rate of interest. If this is
financed via lump-sum taxes, changes in inflation do not change the opportunity cost of holding money a rise in inflation that depreciates the individuals
money holdings is offset by the increase in the transfer the individual anticipates
receiving.

Chapter 3: Money and Transactions


1. Suppose the production function for shopping takes the form = c =
ex (ns )a mb , where a and b are both positive but less than 1 and x is a
1
1
productivity factor. The agents utility is given by v(c, l) = c1 + l1
where l = 1 n ns and n is time spend in market employment.
(a) Derive the transaction time function g(c, m) = ns .
(b) Derive the money in the utility function specification implied by
the shopping production function. How does the marginal utility
of money depend on the parameters a and b? How does it depend on
x?
(c) Is the marginal utility of consumption increasing or decreasing in m?
(a) From the shopping production function,
g(c, m) = ns =

c 1/a
ex mb

(8)

(b) From the definition of the agents utility,


v(c, l) =
=

(1 n ns )1
c1
+
1
1

 c 1/a 1
1

1
ex mb
c
+
u(c, n, m)
1
1

The marginal utility of money is


u(c, n, m)
m



v(c, l) l
=
l
m
  bns 

s 1/a
=
1 n (n )
am

(9)

The time spend shopping can be written as c1/a ex/a mb/a . The marginal productivity of money in reducing shopping time is given by (b/a)(ns /m), so an
increase in b/a increases the effect additional money holdings have in reducing
the time needed for shopping. Additional money holdings result in more leisure
(and more utility) when b/a is large, thus acting to increase the marginal utility
l
rises with b/a. But the marginal utility
of money. In terms of equation (9), m
declines.
of leisure a decreasing function of total leisure, so v(c,l)
l
The effect of x on the marginal utility of money, for given c and n, operates
through ns and represents a productivity shift; an increase in x reduces the time
needed for shopping for given values of c and m. This affects the productivity
of m in the shopping time production function. The marginal product of money
in reducing shopping time is (b/a)c1/a ex/a m(1+b/a) . This is decreasing in x;
8

a higher x decreases the marginal effect of m in reducing shopping time, so


money is less productive.
(c) An increase in consumption affects utility in two ways. First, consumption directly yields utility; vc > 0. This represents the effect of consumption
on utility, holding leisure constant. Since leisure is being held constant, vc is
independent of m. Second, higher consumption increases the time devoted to
shopping, as this reduces the time available for leisure. This effect will depend
on the level of money holding. From (8), consumption and money are complements in producing shopping time, and higher money holdings reduce the effect
of higher c on ns . This means that with higher money holdings, an increase in
c has less of an effect in reducing leisure time and will therefore lead a rise in
consumption to have a larger overall positive effect on utility.
2. Define superneutrality. Carefully explain whether the Cooley-Hansen cashin-advance model exhibits superneutrality. What role does the cash-inadvance constraint play in determining whether superneutrality holds?
A model exhibits the property of superneutrality if the real equilibrium (output, consumption, capital, etc.) is independent of the rate of nominal money
growth. Superneutrality normally is interpreted to refer to the steady-state equilibrium of a model. As demonstrated in Chapter 2, the Sidrauski model displays superneutrality with respect to the steady-state, but changes in the inflation rate will generally affect the short-run equilibrium. If the ratio of the
marginal utilities of leisure and consumption is independent of money holdings,
then Sidrauskis model is superneutral in the short-run also (see pages 65-67).
The Cooley-Hansen model does not display superneutrality. Different rates
of inflation affect the opportunity cost of holding money. Through the cashin-advance constraint, inflation affects the marginal cost of consumption since
consumption is treated as a cash good. Higher inflation induces a substitution
away from cash goods and towards credit goods. In Cooley and Hansens model,
leisure is a credit good; cash is not needed to purchase leisure. As a result,
changes in the steady-state rate of inflation alter the demand for leisure and the
supply of labor. This was shown in equation (3.29) on page 110, where was
equal in the steady-state to one plus the inflation rate.
3. Is the steady-state equilibrium in the Cooley-Hansen cash-in-advance model
affected by any of the following modifications? Explain.
(a) Labor is supplied inelastically (normalize so that n = 1, where n is
the supply of labor).
(b) Purchases of capital are also subject to the cash-in-advance constraint
(i.e. one needs money to purchase both consumption and investment
goods);

(c) The growth rate of money follows the process ut = ut1 + t where
0 < < 1 and is a mean zero i.i.d. process.
(a) Following on the previous problem, one important modification when labor is supplied inelastically is that Cooley and Hansens model will now display
superneutrality. Without a labor-leisure choice, the model becomes essentially
the model of section 3.3.1.
(b) Referring to the model of section 3.6.1, the cash-in-advance constraint
would become
Pt ct + Pt [kt (1 )kt1 ] Mt1 + Tt1
where kt (1 )kt1 is equal to net purchases of capital. Dividing by Pt , this
becomes
mt1
+ t at
(10)
ct + it
t
where it is net investment ( kt (1 )kt1 ).
The value function for this problem is
V (at , kt1 ) = max {u(ct , 1 nt ) + Et V (at+1 , kt )}
t
where at+1 = mt+1
+ t+1 , kt = f(kt1 , nt ) + (1 )kt1 + at ct mt and
the maximization is subject to the cash in advance constraint (10). Let be
the Lagrangian multiplier associated with the budget constraint and let be
the Lagrangian multiplier associated with the cash-in-advance constraint. If we

n1
), then
assume a standard Cobb-Douglas production function ( yt = ezt kt1
t
the budget constraint is

ezt kt1
n1
+ (1 )kt1 + at ct + kt + mt
t

and the first order conditions for ct , kt , mt , and nt , together with the envelope
conditions, are
uc (ct , 1 nt ) t t = 0

(11)

Et Vk (at+1 , kt ) t t = 0

(12)


Et

1
t+1


Va (at+1 , kt ) t = 0


yt
un (ct , 1 nt ) + (1 )
nt


Et Vk (at+1 , kt ) = 0

Va (at , kt1 ) = t + t
10

(13)

(14)

(15)



yt
Vk (at , kt1 ) = t
+ 1 + t (1 )
kt1

(16)

The Lagrangian appears in this last condition because higher capital at the
start of the period reduces the cash needed to achieve a given value of kt ; only
constraint.
net purchases ( kt (1 )kt1 ) are subject to the cash-in-advance

Since (15) implies Et Va (at+1 , kt ) = Et t+1 + t+1 , equations (11) - (16)
can be used to derive the following conditions, which should be compared to
equations (3.51), (3.52), and the two equations following (3.52) on page 126:
uc (ct , 1 nt ) = t + t

Et

t+1 + t+1
t+1

(17)


= t


un (ct , 1 nt ) + (t + t ) (1 )

(18)

yt
nt


=0



t + t = Et Vk (at+1 , kt ) = Et Rt t+1 + t+1 (1 )

(19)

(20)

where Rt = yt+1
kt + 1 . The first two equations are identical to (3.51) and
(3.52). The next two differ. According to (??), the marginal utility of leisure is
set equal to the utility value of the marginal product of labor, but now account
must be taken of the fact that any additional income requires cash to be spent.
That is why the marginal product of labor is multiplied by t + t and not just
t . According the (20), the value of an additional purchase of capital (which
costs t + t ) is the additional future return (the Rt t+1 term) and the value of
relaxing the future cash-in-advance constraint that comes from reducing future
net purchases (the t+1 (1 ) term).
Turning to an analysis of the steady-state, (18) implies that

 ss

1
ss = ss

which implies ss will be requires for the existence of a steady-state since


must be nonnegative. Now eliminate from the steady-state version of (20):


 ss 
 ss

1 (1 )
ss
= Rss ss + ss

ss

or, recalling that Rss 1+ is the steady-state marginal product of capital kyss ,

 ss  

yss
1
1+
ss =
k

which depends on the rate of inflation. Thus, superneutrality does not hold when
capital purchases are also subject to the cash-in-advance constraint. Notice that
11

this conclusion would hold even if labor is supplied inelastically as in part (a)
of this question (see Problem 5 below). By imposing a tax on capital purchases,
inflation affects the steady-state capital stock and kss is decreasing in ss .
For a complete discussion of the implications of making the cash-in-advance
constraint apply to both consumption and capital or only to consumption, see
Abel (1985).
(c) The steady-state depends on the average rate of money growth since that
pins down average inflation. It does not depend on the transitory dynamics of
the monetary supply process, although the short-run dynamics will.
4. Money-in-the-utility-function and cash-in-advance constraints are alternative means for constructing models in which money has positive value in
equilibrium.
(a) What strengths and weaknesses do you see with each of these approaches?
(b) Suppose you wanted to study the effects of the growth of credit cards
on money demand. Which approach would you adopt? Why?
Both the money-in-the-utility function approach and the cash-in-advance approach are best viewed as convenient short-cuts for generating a role for money.
If we believe that the major role money plays is to facilitate transactions, then
in some ways the CIA approach has an advantage in making this transactions
role more explicit. It forces one to think more about the exact nature of the
transactions technology and the timing of payments (e.g., can current period income be used to purchase current period consumption?). On the other hand, the
rather rigid restrictions the CIA typically places on transactions are certainly
unattractive. In modern economies we normally have multiple means that can be
used to facilitate the transactions we undertake. Also, the generally exogenous
distinction between cash and credit goods is troublesome, since most things are
a bit of both.
The MIU approach can be viewed as being based on some specification of
a shopping time model, and the notion of a production function for shopping
time allows for less rigid substitution between different means of carrying out
transactions. The example in the text emphasized the use of time or money
for transactions, but one could allow a variety of means of payment to enter
the production function as imperfect substitutes. Of course that treats the degree of substitution as exogenous, which is also unsatisfactory. We would really
like a model that accounts for why certain means of payment are used in some
circumstances and others in different circumstances.
By emphasizing the link between transactions and money demand, the CIA
approach probably provides the more natural starting point for an analysis of
credit card usage. For an interesting recent analysis, see D.L. Brito and P.R.
Hartley, Consumer Rationality and Credit Cards, Journal of Political Economy,
103 (2), April 1995, 400-433.

12

5. Consider the model of Section 3.3.1. Suppose that money is required to


purchase both consumption and investment goods. The cash-in-advance
constraint then becomes ct + xt mt1 /t + t where x is investment.
.
Assume the aggregate production function takes the form yt = ezt kt n1
t
Show that the steady-state capital-labor ratio is affected by the rate of
inflation. Does a rise in inflation raise or lower the steady-state capitallabor ratio. Explain.
Most of this problem is already worked out as part of the solution to Problem
3.b. The model of Section 3.3.1 assumed utility depended only on consumption,
so there was no labor-leisure choice. Otherwise, the set-up is similar to Problem
3.b, so the equations defining the steady-state are, from (17) - (20),
uc (css ) = ss + ss


ss + ss
ss

(21)


= ss

ss + ss = [Rss ss + ss (1 )]

(22)

(23)

Equation (19) has been dropped since there is no labor supply decision, and
utility in (21) depends only on consumption. From (22),

 ss

1
ss = ss

so (23) becomes


ss

 ss

1 (1 )
= Rss +

For convenience, normalize n to 1. Then Rss + 1 y ss /kss + 1 =


(kss )1 + 1 , and equation (23) implies

 ss 

1
1+
(kss )1 =

Hence, the steady-state capital-labor ratio is


k

ss

 1
   ss 
1

1
1+
=

which is decreasing in the inflation rate ( ss ). Higher inflation implies a higher


tax on capital purchases and this lowers the steady-state stock of capital.

13

6. Consider the following model:


Preferences: Et

i [ln ct+i + b ln dt+i ]

i=0

a
+ t +
Budget constraint: ct + dt + mt + kt = Akt1

Cash-in-advance constraint: ct t +

mt1
+ (1 )kt1
1 + t
(24)

mt1
1 + t

(25)

where m denotes real money balances and t is the inflation rate from
period t 1 to period t. The two consumption goods, c and d, represent
cash (c) and credit (d) goods. The net transfer is viewed as a lump-sum
payment (or tax) by the household.
(a) Does this model exhibit superneutrality? Explain.
(b) What is the rate of inflation that maximizes steady-state utility?
(a) The model exhibits superneutrality if the real variables k, c, and d are
t1
independent of in the steady-state. If we define at t + m
1+ t , the value
function can be defined as
V (at , kt1 ) = max {ln ct+i + b ln dt+i


mt
a
+Et V t+1 +
, Akt1
+ (1 )kt1 + at ct dt mt
1 + t+1
where the maximization is subject to
ct at
Let denote the Lagrangian multiplier associated with this cash-in-advance constraint. From the first order conditions for the agents decision problem,
1
Et Vk (at+1 , kt ) t = 0
ct

(26)

b
Et Vk (at+1 , kt ) = 0
dt

(27)

1
Et Va (at+1 , kt ) Et Vk (at+1 , kt ) = 0
1 + t+1

(28)

Va (at , kt1 ) = Et Vk (at+1 , kt ) + t

(29)

14



a1
Vk (at , kt1 ) = aAkt1
+ 1 Et Vk (at+1 , kt )

(30)

plus the two constraints (24) and (25). Equation (30) implies that, in the steadystate,
1

 1




1
1
ss a1
ss
1 = aA(k )
1+
+1 k =
(31)
A

This means that the steady-state capital stock is independent of the inflation
rate.
Let t Et Vk (at+1 , kt ). From (26) and (27),



dt
= 1+ t b
(32)
ct
t
Equations (28) and (29) imply



t+1 + t+1
= t
Et
1 + t+1

In the steady-state, this implies



ss
ss + ss
1 + ss
=
1
+
=
ss
ss

and combining this with (32),


dss
=
css

1 + ss


b

(33)

so the relative consumption of c and d depends on the rate of inflation. The


real equilibrium does not display superneutrality.
(b) From the steady-state marginal product of capital condition (31), we have
the standard result that the real rate of return will equal 1/. Letting R 1/,
equation (33) can be written as
dss = (1 + iss )bcss
where i = R(1 + ). Letting Z = A(kss )a kss , in the steady-state we have
from the budget constraint (24) Z = css + dss = css + (1 + iss )bcss or css = Z
where = [1 + (1 + iss )b]1 . Hence, steady-state utility of the representative
agent can be expressed as
1
[ln css + b ln dss ] =
1
=
=
=

1
1
1
1
1
1
1+b
1

[ln css + b ln(1 + iss )bcss ]


[ln Z + b ln(1 + iss )bZ]
[(1 + b) ln Z + b ln(1 + iss )b]
ln +
15

1+b
b
ln Z +
ln(1 + iss )b
1
1

Now maximize this with respect to the nominal rate of interest i. Since Z was
shown earlier to be independent of the inflation rate, the first order condition
is




1+b
b
1
b
=0

+
1 1 + (1 + iss )b
1 1 + iss
or
1+b
1
=
1 + iss
1 + (1 + iss )b
which implies
1 + (1 + iss )b
=1+b
1 + iss
This holds if and only if
iss = 0
So the optimal rate of inflation will be the rate that yields a zero nominal rate
of interest.
7. Consider the following model:
Preferences: Et

i (ln ct+i + ln dt+i )

i=0

a
Budget constraint: ct + dt + mt + kt = Akt1
+ t +

mt1
+ (1 )kt1
1 + t

where m denotes real money balances and t is the inflation rate from
period t 1 to period t. Utility depends on the consumption of two types
of good; c must be purchased with cash, while d can be purchased using
either cash or credit. The net transfer is viewed as a lump-sum payment
(or tax) by the household. If a fraction q of d is purchased using cash,
then the household also faces a cash-in-advance constraint that takes the
form
mt1
+ t
ct + qdt
1 + t
What is the relationship between the nominal rate of interest and whether
the cash-in-advance constraint is binding? Explain. Will the household
ever use cash to purchase d (i.e. will the optimal q ever be greater than
zero)?

16

The basic model is similar to the one studied in Problem 6, differing only in
the utility function and the cash-in-advance constraint. The value function is
V (at , kt1 ) = max {ln ct+i + ln dt+i


mt
a
+Et V t+1 +
, Akt1
+ (1 )kt1 + at ct dt mt
1 + t+1
t1
where at = t + m
1+ t and ct + qdt at . and we require that 0 q 1
since q is the fraction of the d good purchased with cash. Actually, the relevant
consideration is whether q is positive or not. Let denote the Lagrangian on
the constraint q 0. The first order conditions for the households decision
problem for the current are simply stated here as, modifying them to reflect the
different utility function and cash-in-advance constraint:

1
Et Vk (at+1 , kt ) t = 0
ct
1
Et Vk (at+1 , kt ) qt = 0
dt
1
Et Va (at+1 , kt ) Et Vk (at+1 , kt ) = 0
1 + t+1
Va (at , kt1 ) = Et Vk (at+1 , kt ) + t


a1
Vk (at , kt1 ) = aAkt1
+ 1 Et Vk (at+1 , kt )
In addition, we need the first order condition for the optimal choice of q. This
takes the form
t dt + t 0 qt t = 0
where the condition qt t = 0 is the complementary slackness condition associated
with the inequality constraint on q. Since q cannot be reduced below zero, the
optimum can have t dt +t < 0 at q = 0; utility could be increased by reducing
q even further, but the non-negativity constraint binds. As long as the nominal
rate of interest is positive, > 0, and d > 0. this implies that > 0 from
which the condition q = 0 implies that q = 0. So, as long as the nominal rate
of interest is positive, the household will never use cash to purchase d.
8. Trejos and Wright (1993) find that if no search is allowed while bargaining
takes place, output tends to be too low (the marginal utility of output
exceeds the marginal production costs). Show that output is also too low
in a basic cash-in-advance model. (For simplicity, assume only labor is
needed to produce output according to the production function y = n.)
Does the same hold true in a money-in-the-utility-function model?

17

In a basic cash-in-advance model, inflation taxes cash goods. Suppose the


nominal rate of interest is positive; relative to the case of a zero nominal interest
rate, households will be consuming fewer cash goods (which bear the inflation
tax) and more credit goods. Since leisure is a credit good, inflation will tend to
lower output by increasing the demand for leisure and reducing labor supply. For
example, equation (3.29) on page 110 shows how inflation reduces labor supply
relative to the case of a zero nominal rate of interest.
If we modify the model of Section 3.3.2.1 by ignoring capital, and assume the
production function is y = n, then the value function for the decision problem
of the household becomes (see section 3.6 of the Chapter Appendix):



nt + at ct
+ t+1
V (at ) = max u(ct , 1 nt ) + V
t+1
where at = t + mt1 /t , and the maximization is subjective to the cash-inadvance constraint ct at . If t is the Lagrangian multiplier associated with
the cash-in-advance constraint, then the first order necessary conditions are
uc (ct , 1 nt )

V  (nt + at ct mt )
t = 0
t+1

ul (ct , 1 nt ) +

V  (at ) =
Let t

V  (nt +at ct mt )
.
t+1

V  (nt + at ct mt )
=0
t+1

V  (nt + at ct mt )
+ t
t+1

Then these first order conditions imply


1
t
ul (ct , 1 nt )
t
=
= 1+
1
uc (ct , 1 nt )
t + t
t
As long as the cash-in-advance constraint is binding, > 0 and ul /uc is greater
than it would be in the case in which = 0. Since ul /uc is increasing in labor
supply, labor supply and output is reduced relative to the = 0 case. In this
framework, the marginal cost of output is ul since this is the utility cost of
supplying additional labor. The marginal utility of the output that is produced
is uc . Since ul < uc when the cash-in-advance constraint binds, the marginal
utility of output exceeds the marginal cost of production.
In a basic money-in-the-utility-function model, the relevant condition was
given by equation (2.34) on page 66. The marginal utility cost of supplying
more labor ul is just equal to the marginal utility of consumption times that
additional output produced fn uc . So the marginal cost of production and the
marginal utility of output are equal. This doesnt mean money and inflation
dont affect output. A positive nominal interest rate reduces real money holdings
18

relative to the social optimum. How that affects labor supply (and output) will
depend on how a decrease in m affects ul /uc and the effect could go either way.
For the utility function used in the linear version of the money-in-the-utilityfunction model of Chapter 2, equation (2.45) shows that a lower value of m will,
for given c and y, act to increases labor supply for < 1 and decrease labor
supply for > 1. Thus, if, for example, < 1, consumption and money are
complements; an increase in m increases the marginal utility of consumption.
Higher inflation that reduces m also leads to a fall in the marginal utility of
consumption. Households will shift towards consuming more leisure and fewer
consumption goods. The decline in labor supply as more leisure is consumed will
lower output.

19

Chapter 4: Money and Public Finance


1. Consider the version of the Sidrauski (1967) model studied in Problem
1 of Chapter 2. Utility was given by u(ct , mt ) = w(ct ) + v(mt ), with
w(ct ) = ln ct and v(mt ) = mt (B D ln mt ) where B and D are positive
parameters. Steady-state revenue from seigniorage is given by m, where
is the growth rate of the money supply.
(a) Is there a Laffer Curve for seigniorage (i.e. are revenues increasing
in for all and decreasing in for all > for some ?
(b) What rate of money growth maximizes steady-state revenues from
seigniorage?
(c) Assume now that the economys rate of population growth is n and
reinterpret m as real money balances per capita. What rate of inflation maximizes seigniorage? How does it depend on n?

(a) From Problem Set 2, we know that the demand for money in this model
B
1. Hence, seigniorage in the
is given by mt = Aei/ct D where ln A = D
steady-state is equal to
sss = Ae(r

ss

+)/css D

Taking the derivative with respect to ,



ss
ss
ss
ss
ss
ss


sss
= Ae(r +)/c D ss Ae(r +)/c D = Ae(r +)/c D 1 ss

c D
c D
This is positive (i.e. seigniorage is increasing in inflation) for < css D, and
negative for > css D. Hence, there is a Laffer curve.
(b) Steady-state seigniorage is maximized for = = css D.
(c) With population growth at the rate , the growth rate of per capital money
balances is given by

Hence, in the steady-state, = . Steady-state seigniorage will be still be


maximized at an inflation rate of css D, but this now corresponds to a rate of
money growth of css D + .
2. Suppose that government faces the following budget identity:
bt = Rbt1 + gt t yt st
where the terms are one period debt, gross interest payments, government purchases, income tax receipts and seigniorage. Assume seigniorage
20

is given by f (t ) where is the rate of inflation. The interest factor

R is constant and the expenditure process {gt+i }i=0 is exogenous. The


government sets time paths for the income tax rate and for inflation to
minimize
Et

i [h( t+i ) + k(t+i )]

i=0

where the functions h and k represent the distortionary costs of the two
tax sources. Assume the functions h and k imply positive and increasing
marginal costs of both revenue sources.
(a) What is the intratemporal optimality condition linking the choices of
and at each point in time?
(b) What is the intratemporal optimality condition linking the choice
at different points in time?
(c) Suppose y = 1, f () = a, h( ) = b 2 and k() = c2 . Evaluate the
inter- and intratemporal conditions.
 i Find the optimal settings for t
and t in terms of bt1 and
R gt+i .
(d) Using your results from part (c), when will optimal financing imply
constant planned tax rates and inflation over time?
(a) Solving the budget 
constraint forward, the governments decision problem
i
can be written as min Et
i=0 (h( t+i ) + k( t+i )) subject to


Rbt1 + Et
Ri gt+i Et
Ri [ t+i yt+i + f ( t+i )] = 0
Let be the Lagrangian multiplier associated with this constraint. The first
order conditions are


Et i h ( t+i ) Ri yt+i = 0
and


Et i k (t+i ) Ri f  ( t+i ) = 0
Hence, the condition linking taxes and seigniorage at each date t + i take the
form

Et k (t+i )
Et h ( t+i )
=
=
Et yt+i
(R)i
Et f  (t+i )
(b) The first order conditions for seigniorage at dates t + i and t + j take
the form i Et k (st+i ) = Ri Et f  (t+i ) and j Et k (st+j ) = Rj Et f  (t+j ),
or
Et (R)i

k (t+i )
k (t+j )
= = Et (R)j 

f ( t+i )
f (t+j )
21

(c) Given the assumed functional forms, the first order conditions become
2bEt t+i =

c

Et t+i
=
2
(R)i
a

c
Et t+i for all i. The intertemporal condition
which implies that Et t+i = ab
becomes
c
c
Et t+i (R)i = 2
Et t+j (R)j or for j = 0, Et t+i = (R)i t
2
a
a
c
These results imply Et t+i = ab
(R)i t .
Now we can evaluate the governments budget constraint recalling at y = 1):


Rbt1 + Et
Ri gt+i = Et
Ri ( t+i yt+i + f (t+i ))

c
=
Ri ( (R)i t + a(R)i t )
ab

c
+ a t
=
(R2 )i
ab

This implies that


t =


1 
c

+a
B Rbt1 + Et
Ri gt+i
ab

where B = R2 /(R2 1).


Finally, the optimal tax rate is given by
t =


1 
c
 c  c

t =
+a
B Rbt1 + Et
Ri gt+i
ab
ab
ab

(d) Et t+i = (R)i t = t if and only if R = 1 (i.e. R = 1/).


3. Mankiw (1987) suggested that the nominal interest rate should evolve as a
random walk under an optimal tax policy. Suppose the real rate of interest
is constant and that the equilibrium price level is given by equation (4.24).
Suppose the nominal money supply is given by mt = mpt + vt where mpt is
the central banks planned money supply and vt is a white noise control
error. Let be the optimal rate of inflation. There are different processes
for mp that lead to the same average inflation rate but different time series
behavior of the nominal interest rate. For each of the processes for mpt
given in a and b, show that average inflation is ; also show whether the
nominal interest rate is a random walk.
(a) mpt = (1 )t + mt1 ;
(b) mpt = mt1 + .

22

Equation (4.24) states that


pt =

Et pt+1
mt
+
1+
1+

(34)

For the money process in part (a), this becomes


pt =

(1 )t + mt1 + vt Et pt+1
+
1+
1+

(35)

and the no-bubbles solution is of the form


pt = p0 + at + bmt1 + cvt
where a, b, and c are coefficients to be determined. This solution implies
Et pt+1 = p0 + a(t + 1) + bmt = p0 + a(t + 1) + b [(1 )t + mt1 + vt ]
Using this and the trial solution in equation (35) yields
p0 + at + bmt1 + cvt

(1 )t + mt1 + vt
1+
[p0 + a(t + 1) + b ((1 )t + mt1 + vt )]
+
1+


(1 ) + a + b(1 )
(p0 + a)
=
t+
1+
1+




(1 + b)
1 + b
+
mt1 +
vt
1+
1+

This will hold for all realizations of vt and mt1 if


p0 =

a=

(p0 + a)
p0 = a
1+

(1 ) + a + b(1 )
1+
b=


a = (1 )(1 + b)

(1 + b)
b=
1+
1 + (1 )
c=

1
1 + (1 )

Substituting for b in the expression for a,







1+
a = (1 ) 1 +
= (1 )
1 + (1 )
1 + (1 )
23

Hence, the equilibrium price level evolves according to






1+
1
pt = (1 )
( + t) +
(mt1 + vt )
1 + (1 )
1 + (1 )
Average inflation will equal
pt


1+
mt1
+
1 + (1 )
1 + (1 )


1+

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

= (1 )

Expected inflation is equal to



 

1+
1
Et pt+1 pt = (1 )
+
(mt mt1 vt )
1 + (1 )
1 + (1 )

 

1+
1
= (1 )
+
(mt vt )
1 + (1 )
1 + (1 )
With a constant real rate of interest, as was assumed in deriving equation (4.24),
the nominal rate of interest will equal

 

1+
1
it = r0 + (1 )
+
(mt vt )
1 + (1 )
1 + (1 )


1
= i0 +
(mt vt )
1 + (1 )
Since mt = (1)+mt1 +vt , mt is a first order autoregressive process
and (1 L)mt = (1 ) + vt . So the nominal interest rate is of the form
it = i0 + zt vt
where zt AR(1) and vt is white noise. Quasi-first differencing,
(1 L)it

= (1 )i0 + (1 L)(zt vt )
= (1 )i0 + (1 ) + vt (1 L)vt
= (1 )i0 + (1 ) + vt1

Thus, the nominal interest rate, so the nominal interest rate will follow the first
order autoregressive process
it = i0 + it1 + vt1

With the money supply process in (b), equation (4.24) becomes


pt =

+ mt1 + vt Et pt+1
+
1+
1+
24

(36)

and the equilibrium solution for pt is of the form pt = p0 + bmt1 + cvt . Using
this in (36),
p0 + bmt1 + cvt

+ mt1 + vt [p0 + b( + mt1 )]


+
1+
1+






1 + b
1
1 + b
p0
+
=
+
mt1 +
vt
1+
1+
1+
1+
=

or p0 = (1 + ), b = 1, and c = 1/(1 + ). With


pt = (1 + ) + mt1 +

1
vt
1+

average inflation is just . Expected inflation is


Et pt+1 pt = mt mt1 +

1
vt = +
1+


1
vt
1+

With a constant real rate of interest, as was assumed in deriving equation


(4.24), the nominal rate of interest will equal


1
it = r0 + +
vt
1+
so that it is equal to a constant plus a white noise error; it is not a random
walk.
4. Suppose the Correia-Teles model of Section 4.5.3 is modified so that output
is equal to f(n) where f is a standard neoclassical production function
exhibiting positive but diminishing marginal productivity of n. If f (n) =
na for 0 < a < 1, does the optimality condition given by (4.44) continues
to hold?
Start with the budget constraint when f(n) = na . Equation (4.42) becomes
dt1



Di ct+i (1 t+i )f (1 lt+i nst+i ) + Rt It+i mt1+i

i=0



Di ct+i (1 t+i )(1 lt+i nst+i )a + Rt It+i mt1+i

i=0

while the first order condition for labor supply, equation (4.39), is modified to
become
ul
= at (1 lt nst )a1
(1 t )

25

(37)

Following the same steps outlines on page 170, we use the fact that dt1 = 0
and Di = i D0 t+i /t to obtain


D0 i 
t+i ct+i t+i (1 t+i )(1 lt+i nst+i )a t+i It+1+i mt+i = 0
t i=0

From (37), t+i (1 t+i ) = ul (1lt nst )1a /a. Making this substitution (along
with the others discussed in the text) yields
0 =
=

ul (1 lt nst )1a
D0 i
uc ct+i ul gnst+i
(1 lt+i nst+i )a
t i=0
a

u 

D0 i 
l
(1 lt nst )
uc ct+i ul gnst+i
t i=0
a

This implies




u 
1
l
s
(1 lt ) ul
nt+i = 0
uc ct+i
a
a
i=0

which corresponds to equation (4.43) on page 171. Notice that the only modification is that ns is multiplied by the factor a1 ; in the example of the text,
a = 1 and this became 1 .
The first-order condition for the optimal choice of m in the social welfare
problem is


1
i ul ( ) t+i g = 0
a
which replaces (4.44). As long as i ul ( a1 ) t+i must be nonzero, the
optimum still involves g = 0, or a zero nominal interest rate.

26

Chapter 5: Money and Output in the Short


Run



b 1

c( M
P )

1. Assume household preferences are given by U =


+ (1N)
1
1
and aggregate output is given by Y = K N 1 . Linearize around the
steady-state the labor market equilibrium condition equation (5.26) from
the monopolistic competition model. How does the result depend on q?
Explain.
Equation (5.26) states that
Ul
= qM P L
Uc
For the functional forms specified in the question, this becomes
Y
(1 N)
= q(1 )
c (mb )1
N

(38)

where m = M/P .
Using the methods employed in the text, we can define X ss as the steadystate value of X and x as the percent deviation of x around its steady-state
value, so x = xss (1 + x). Then, (38) can be written as


Lss (1 + l)
Y ss (1 + y)
(39)

1 = q(1 ) ss
N (1 + n
)

b
ss
ss
(m (1 + m))
[c (1 + c)]

where L = 1 N. Since
Y ss
(Lss )

1 = q(1 ) ss
N
(css ) (mss )b
equation (39) becomes
(1 + l)
(1 +

c) [(1

1
m)
b]

(1 + y)
(1 + n
)

(40)

Notice that q has dropped out; the labor market deviations around the steadystate will not be affected by q. Now take logs of (40) and employ the approximation that ln(1 + x) x for small x, to obtain
l +
c b(1 )m
= y n

ss
), this implies
l = N
. Using this we have,
Since Lss (1 + l) = 1 N ss (1 + n
Lss n


N ss
ss n
+
c b(1 )m
= y n

27

or


n
=

1
ss
1 + N
Lss


(
y
c + b(1 )m)

2. The Chari, Kehoe, and McGratten (1996) model of price adjustment led
to equation (5.30). Using equation (5.29), show that the parameter a in

(5.30) equals (1 )/(1 + ).


Equation (5.29), page 200, states that




1 1

pt =
[
pt1 + Et pt+1 ] +
[mt + Et mt+1 ]
2 1+
1+

(41)

If m follows a random walk as was assumed in deriving (5.30), Et mt+1 = mt


and (5.29) becomes




1 1
2
pt =
[
pt1 + Et pt+1 ] +
mt
2 1+
1+


and rewrite this as
It will be convenient to define b 12 1
1+
pt = b [
pt1 + Et pt+1 ] + (1 2b) mt

(42)

Let the proposed solution be


pt = a1 pt1 + a2 mt
From (5.30), Et pt+1 = a1 pt + a2 mt = a1 (a1 pt1 + a2 mt ) + a2 mt , so equation
(42) becomes
pt

= b [
pt1 + a1 (a1 pt1 + a2 mt ) + a2 mt ] + (1 2b) mt


= b 1 + a21 pt1 + [1 2b + ba2 (1 + a1 )] mt

For this to equal the proposed solution for all realizations of pt1 and mt requires
that


a1 = b 1 + a21
and
a2 = 1 2b + ba2 (1 + a1 ) a2 =

1 2b
1 b(1 + a1 )

The first of these conditions requires that a1 be the solution to


 
1
a21
a1 + 1 = 0
b
28

or
a1 =
Recalling that b was equal to

a1

2


1+
1

1
2

b1

1
1+

b2 4
2


, this becomes

 
2
1+
4 1
4
2



2
1+
=
1

1

 

1+
1
=
1 + 2 + 2 (1 2 + 2 )

1
1






2
2
1+
1
1+
1

=
and
=
2
1
1
1
1

Since 1 = (1 + )(1 ), we can write these two potential solutions as



2

1
1
=

1
1+
1+
1

and


2

1+
1+
=

1
1

Both only the first of these is less than 1 in absolute value, so the stable solution
has

1
a1 =

1+
Returning to the condition for a2 , and using the value for a1 just found,
together with the definition of b,


1
1

1+
1 2b


=
a2 =

1
1 1
1 b(1 + a1 )
1+
1
2

=
=

2
1+

1+

1+

2


1

1
1 + 1+



1
2
=1

+
1+


1
1+



1
1+

which verifies that a2 = 1 a1 .

29

=

3. Equation (5.28) was obtained from equation (5.27) by assuming R = 1.


Show that in general,






Rss
1
1
Rss
pt =
pt + ss Et pt+1 +
vt + ss Et vt+1
1 + Rss
R
1 + Rss
R

Equation (5.27) of the text states that





Et Pt Vt Yt + R1
t+1 Pt+1 Vt+1 Yt+1

Pt =
1
1q
1
qEt Pt1q Yt + Rt+1
Pt+1
Yt+1
If we evaluate this at the steady state, recalling that = (2 q)/(1 q),


(P ss ) V ss Y ss 1 + (Rss )1
1

 = P ss V ss
P ss =
1
1
q
ss
ss
ss
1q
1 + (R )
q (P )
Y

(43)

(44)

Let lower case letters denote percentage deviations from the steady-state, so
1+ x = Xt /X ss . Then the left side of equation (5.27) can be written P ss (1+ pt )
while the right side becomes



1
(P ss ) V ss Y ss (1 + pt ) (1 + vt )(1 + yt ) + (Rss ) Et (1 + rt+1 )1 (1 + pt+1 ) (1 + vt+1 )(1 + yt+1 )


1
1
1
(P ss ) 1q qY ss (1 + pt ) 1q (1 + yt ) + (Rss )1 Et (1 + rt+1 )1 (1 + pt+1 ) 1q (1 + yt+1 )
Now using the approximations (1 + x)s 1 + sx and (1 + x)(1 + z) 1 + x + z,
this becomes


P ss V ss (1 + pt + vt + yt ) + (Rss )1 Et (1 + pt+1 + vt+1 + yt+1 rt+1 )








1
1
pt + yt + (Rss )1 Et 1 + 1q
pt+1 + yt+1 rt+1
q 1 + 1q
After some cancellation, equation (43) can be written


1
(1 + pt + vt + yt ) + (Rss ) Et (1 + pt+1 + vt+1 + yt+1 rt+1 )







(1 + pt ) = 
1
1
1 + 1q
pt + yt + (Rss )1 Et 1 + 1q
pt+1 + yt+1 rt+1
(45)
This expression is of the form
1+x=

1 + z + R1 (1 + d)
1 + s + R1 (1 + c)
30

(46)

which can be written as




(1 + x) 1 + s + R1 (1 + c) = 1 + z + R1 (1 + d)
Multiplying out the left side,
1 + x + s + sx + R1 (1 + x) + R1 (x + xc) 1 + x + s + R1 (1 + c) + R1 x


= 1 + R1 x + 1 + s + R1 (1 + c)
yielding for equation (46),


1 + R1 x + 1 + s + R1 (1 + c) 1 + z + R1 (1 + d)
or
x

z + R1 (1 + d) s R1 (1 + c)
=
(1 + R1 )

R
1+R


zs+

 

1
(d c)
R



1
pt + y t ,
Returning to equation (45), we have x = pt , z = pt + vt + yt , s = 1q



1
pt+1 + yt+1 rt+1 , so
d = Et (pt+1 + vt+1 + yt+1 rt+1 ), and d = Et 1q

pt =
since

1
1q

Rss
1 + Rss



1
(p
+
v
)
pt + vt +
E
t
t+1
t+1
Rss

= 1. This is our desired result.

4. Using the equilibrium condition (5.42) for the price level, show that equilibrium output is independent of any policy response to t1 or vt1 .
Equation (5.42) states that
pt =

d2 mt + a(1 + d2 )Et1 pt + Et pt+1 d2 vt + ut (1 + d2 )t


(1 + a)(1 + d2 )

(47)

Suppose that monetary policy does respond to t1 and vt1 by setting the nominal supply of money according to
mt = b1 t1 + b2 vt1
Substituting this expressing into (47),
pt =

d2 (b1 t1 + b2 vt1 ) + a(1 + d2 )Et1 pt + Et pt+1 d2 vt + ut (1 + d2 )t


(1 + a)(1 + d2 )
(48)

31

Assuming all the disturbance terms are serially uncorrelated, we can use the
method of undetermined coefficients to find the solution for the equilibrium price
level. Inspection of (48) suggests the following guess:
pt = 0 + 1 t1 + 2 vt1 + 3 vt + 4 ut + 5 t

(49)

where xt d2 vt + ut (1 + d2 )t . Using (49),


Et1 pt = 0 + 1 t1 + 2 vt1

(50)

Et pt+1 = 0 + 1 t + 2 vt

(51)

and

Substituting equations (49) - (51) into (48), yields, with some rearranging,
(1 + a)(1 + d2 )pt

= d2 (b1 t1 + b2 vt1 ) + a(1 + d2 ) ( 0 + 1 t1 + 2 vt1 )


+ 0 + 1 t + 2 vt d2 vt + ut (1 + d2 )t
= 0 [1 + a(1 + d2 )] + [d2 b1 + a 1 (1 + d2 )] t1
+ [d2 b2 + 2 a(1 + d2 )] vt1
+ [ 1 (1 + d2 )] t + [ 2 d2 ] vt + ut
(52)

We could now replace pt on the left side of this equation with the proposed
solution given in equation (49) and equate coefficients to solve for the values of
the i coefficients. However, to determine the effect of the policy reaction on
output, we do not need to do this. From equation (5.34) on page 205, output
depends on the price surprise term pt Et1 pt . We can obtain this by taking
expectations of (52) based on t1 information and subtract the result from (52).
So first taking expectations,
(1 + a)(1 + d2 )Et1 pt

= 0 [1 + a(1 + d2 )]
+ [d2 b1 + a 1 (1 + d2 )] t1 + [d2 b2 + 2 a(1 + d2 )] vt1

Subtracting this from (52),


(1 + a)(1 + d2 ) (pt Et1 pt ) = [ 1 (1 + d2 )] t + [ 2 d2 ] vt + ut
This is independent of the policy response coefficients b1 and b2 .
Because any systematic policy response to t1 or vt1 is fully incorporated
into the publics expectations at the start of period t, it cannot generate any price
surprise; pt (and Et1 pt ) adjust fully to anticipated or predictable movements
in the period t nominal supply of money.
5. Assume nominal wages are set for one period but that they can be indexed
to the price level:
wtc = wt0 + b(pt Et1 pt )
where w0 is a base wage and b is the indexation parameter (0 b 1).
32

(a) How does this change modify the aggregate supply equation given by
(5.18)?
)2 .
(b) Assume the indexation parameter is set to minimize Et1 (nt n
Using your modified aggregate supply equation, together with (5.35)
- (5.37) and a money supply process mt = t , show that the optimal degree of wage indexation is increasing in the variance of and
decreasing in the variance of (Gray 1978).
(a) From equation (5.16) and the new specification for the contract wage,
employment is given by


nt = yt wt0 + b(pt Et1 pt ) + pt
If the base wage is set according to (5.15), wt0 = Et1 (yt + pt nt ), and
nt Et1 nt = yt Et1 yt + (1 b) (pt Et1 pt )

(53)

Notice that if b = 0 (no indexation), we obtain the expression in the test (equation 5.16). At the other extreme, if b = 1, nominal wages are completely indexed
and adjust fully to unexpected changes in the price level. As a result, the real
wage and employment are insulated from price level movements.
Since the model underlying equations (5.34) - (5.37) was based on the assumption that labor supply was inelastic, we can set Et1 nt = 0 since all variables should be interpreted as deviations around a steady-state.
Substituting (53) into the production function (5.8),
yt Et1 yt

= (1 ) [yt Et1 yt + (1 b) (pt Et1 pt )] + zt Et1 zt


= a(1 b) (pt Et1 pt ) + t
(54)

where, as in the text, a = (1 )/ and t = (zt Et1 zt ) /. Equation (54)


shows that, relative to (5.18), the effect of a price surprise on output is now
a(1 b) < a.
(b) To solve for the variance of employment, use (5.35) - (5.37) to solve for
yt and pt , using the assumption that mt = t . From (5.36) and (5.37),
rt

= it + pt Et1 pt
= d2 (yt + pt + vt t ) + pt Et1 pt

Substituting this into the aggregate spending equation (5.35),


yt

= Et yt+1 [d2 (yt + pt + vt t ) + pt Et1 pt ] + ut


Et yt+1 [(1 + d2 )pt Et1 pt ] + ut + d2 (t vt )
=
1 + d2

(55)

If we assume the productivity shock z is serially uncorrelated, then Et yt+1 =


Et zt+1 = 0 and (55) implies
yt Et1 yt = (pt Et1 pt ) + st
33

(56)

where st [ut + d2 (t vt )] /(1 + d2 ). Solving (54) and (56) for yt Et1 yt


and pt Et1 pt ,


a(1 b)
1
t
yt Et1 y =
st +
1 + a(1 b)
1 + a(1 b)

pt Et1 pt =

st t
1 + a(1 b)

Using these results in (53),


nt Et1 nt

= yt Et1 yt + (1 b) (pt Et1 pt )






(1 b)(1 + a)
b
=
st +
t
1 + a(1 b)
1 + a(1 b)

(57)

so
2

)
Et1 (nt n

2


b
(1 b)(1 + a)
= Et1

st +
t n
1 + a(1 b)
1 + a(1 b)


2
2
(1 b)(1 + a)
b
=
2s +
2 + n
2
1 + a(1 b)
1 + a(1 b)


The value of the indexation parameter is picked to minimize this expression.


The first order condition for this problem is




(1 + a)b
(1 b)(1 + a)2
2
s + 2
2 = 0
2
3
3
[1 + a(1 b)]
[1 + a(1 b)]
which implies
(1 b)(1 + a) 2s + b2 = 0
or the optimal degree of indexation is
b =

(1 + a)2s
2
=1
2
2
(1 + a)s +
(1 + a) 2s + 2

Hence,
0 b 1
with the inequalities strict if both 2 and 2s are positive. Define 2 /2s as
the relative variance of productivity shocks to demand side shocks (arising from
money supply shocks , money demand shocks v, and aggregate spending shocks
u). Then
b = 1

1+a+

34

which is decreasing in . As aggregate demand shocks become more important


(and falls), it is optimal to have a higher degree of nominal wage indexation to
insulate real wages and employment from fluctuating. Real productivity shocks
do call for real wage adjustments, so if shocks are important, then the optimal
degree of indexation is lower in order to allow for some real wage movements
as the price level changes.
6. The basic Taylor model of price level adjustment was derived under the
assumption that the nominal wage set in period-t remained unchanged for
periods t and t + 1. Suppose instead each period t contract specifies a
nominal wage x1t for period t and x2t for period t + 1. Assume these are
given by x1t = pt + yt and x2t = Et pt+1 + Et yt+1 . The aggregate price
level at time t is equal to pt = 12 (x1t + x2t1 ). If aggregate demand is given
by yt = mt pt and mt = m0 + t , what is the effect of a money shock
t on pt and yt ? Explain why output shows no persistence after a money
shock.
From the definition of the aggregate price level and the contract nominal
wages,
pt

1
[pt + yt + Et1 pt + Et1 yt ]
2
= yt + Et1 pt + Et1 yt

(58)

which can be compared to the equation at the bottom of page 216. Notice that
pt1 does not appear in (58), since x2t1 is now set based on Et1 pt rather than
on pt1 and Et1 pt as in the specification given by equation (5.44).
Substituting the assumed specification for aggregate demand into (58),
pt

= (mt pt ) + Et1 pt + Et1 (mt pt )


mt + (1 )Et1 pt + m0
=
1+

(59)

where use has been made of the fact that Et1 mt = m0 under the assumed
money supply process. We can write the solution to this as
pt = 0 + 1 t
for 0 and 1 such that
0 + 1 t =

(m0 + t ) + (1 ) 0 + m0
1+

or
1 =

1+

35

and
0 = m0
Aggregate output is then given by

yt = mt pt = m0 + (1 1 ) t 0 =

1
1+


t

Since output is equal to the white noise error , it displays no persistence.


In the standard formulation, some nominal wages in effect during period t
were set in earlier periods on the basis of the price level in those earlier periods.
This imparts sluggishness to the adjustment of the price level; pt can no longer
jump to fully offset any change in the period t nominal supply of money. With
the alternative specification used in this problem, nominal wages in effect in
period t depend only on period t prices and previous expectations about pt . Thus,
output in period t is only affected by movements in mt that were unpredictable
when the oldest contract still in effect was set.
7. The p-bar model led to the following two equations for pt and yt :
pt =

yt =

d2 mt + Et pt+1 d2 vt + ut
yt
1 + d2

d2 (mt Et1 mt ) + Et pt+1 Et1 pt+1 d2 vt + ut


+ (1 )yt1
1 + d2

Assume = 1 and
mt = mt1 + aut
Show that the variance of yt depends on the parameter a. What value of
a would minimize the impact of IS shocks (u) on output?
If = 1, lagged output drops out of the model, and from the assumed process
for money, mt Et1 mt = mt1 + aut mt1 = aut . Thus, the output equation
can be written as
yt =

Et pt+1 Et1 pt+1 d2 vt + (1 + ad2 ) ut


1 + d2

(60)

Notice that as long as neither Et pt+1 nor Et1 pt+1 are affected by an IS shock,
the impact of u on output would be neutralized if a were set equal to 1/d2 . To
check whether price expectations are affected, use the price and output equations,
together with the money supply process to obtain
(1 + d2 )pt

= d2 (mt1 + aut ) + Et pt+1 d2 vt + ut


[Et pt+1 Et1 pt+1 d2 vt + (1 + ad2 ) ut ]
36

(61)

or
(1 + d2 )pt = d2 (mt1 + aut ) + Et1 pt+1 ad2 ut

(62)

Consider the following proposed solution for pt :


pt = 0 + 1 mt1 + 2 ut
Based on this solution,
pt+1

= 0 + 1 mt + 2 ut+1
= 0 + 1 (mt1 + aut ) + 2 ut+1

So
Et pt+1 = 0 + 1 (mt1 + aut )
and
Et1 pt+1 = 0 + 1 mt1
Substituting these expressions into (62),
(1 + d2 )pt

= d2 (mt1 + aut ) + Et1 pt+1 ad2 ut


= d2 (mt1 + aut ) + 0 + 1 mt1 ad2 ut
= 0 + ( 1 + d2 ) mt1

Using the proposed solution to eliminate pt ,


(1 + d2 ) [ 0 + 1 mt1 + 2 ut ] = 0 + ( 1 + d2 ) mt1
equating coefficients implies
(1 + d2 ) 0 = 0 0 = 0
(1 + d2 ) 1 = ( 1 + d2 ) 1 = 1
2 = 0
so pt = mt1 .
We can now collect these results to evaluate equation (60) for output:
yt =

Et pt+1 Et1 pt+1 d2 vt + (1 + ad2 ) ut


[1 + a(1 + d2 )] ut d2 vt
=
1 + d2
1 + d2

and the variance of output is




2
2
1 + a(1 + d2 )
d2
2u +
2v
2y =
1 + d2
1 + d2
37

Thus, to insulate output from demand shock, a should be set equal to


a =

1
<0
1 + d2

Notice that this is a smaller response than found earlier when the possible effects
of u on expected future prices were ignored (see equation 60). A positive ut that
results in a fall in mt causes private agents to revise downward their forecast
of the future price level: Et pt+1 Et1 pt+1 = aut if a is negative. But from
(60), this acts to reduce yt . Thus, to stabilize yt , the reduction in mt needs to
be smaller than would be the case if price expectations did not matter.
8. Derive the equilibrium expression for pt and yt corresponding to equations (5.39) and (5.41) for the case in which the aggregate productivity
disturbance is given by zt = zt1 + et , 1 < < 1.
The equations of the model that led to equations (5.39) and (5.40) were given
by (5.34) - (5.37) and are repeated here:
yt = Et1 yt + a(pt Et1 pt ) + t

(63)

yt = Et yt+1 rt + ut

(64)

mt pt = yt d1
2 it + vt

(65)

it = rt + Et pt+1 pt = rt + Et t+1

(66)

Aggregate supply:

Aggregate demand:
Money demand:
Fisher equation:

As discussed in the text (page 208), equations (64) - (66) can be combined to
yield equation (5.38) of the text:
yt =

d2 (mt pt ) + Et t+1 + Et yt+1 d2 vt + ut


1 + d2

As discussed on page 204 (and in footnote 43),


Et yt+1 = Et zt+1
Using the assumed process for z specified in the question,
Et yt+1 = zt
Equation (67) then becomes
yt =

d2 (mt pt ) + Et t+1 + zt d2 vt + ut
1 + d2
38

(67)

Equating this expression for yt with the expression for yt given by the aggregate
supply relationship (63) and solving for pt (and using the fact that zt1 + et =
zt ) results in
pt =

d2 mt + a(1 + d2 )Et1 pt + Et t+1 + zt d2 vt + ut (1 + d2 ) (zt1 + t )


d2 + a(1 + d2 )

which corresponds to (5.39).


Taking expectations of this expression as of time t 1 and subtracting the
result from pt yields
pt Et1 pt =

d2 (mt Et1 mt ) + Et t+1 Et1 t+1 + et d2 vt + ut (1 + d2 )t


d2 + a(1 + d2 )

so that real output, from (63) is


yt

= Et1 yt + t


d2 (mt Et1 mt ) + Et t+1 Et1 t+1 + et d2 vt + ut (1 + d2 )t
+a
d2 + a(1 + d2 )


d2
= zt1 +
t
d2 + a(1 + d2 )
a [d2 (mt Et1 mt ) + Et t+1 Et1 t+1 + et d2 vt + ut ]
+
d2 + a(1 + d2 )

where the only difference from equation (5.41) is the presence of et . This is the
innovation in zt that persists into period t + 1. This affects aggregate demand
in period t under the assumption that agents are forward looking.
9. Suppose that the nominal money supply evolves according to mt = +
mt1 + t for 0 < < 1 and t a white noise control error. If the rest
of the economy is characterized by equations (5.34) - (5.37), solve for the
equilibrium expressions for the price level, output, and the nominal rate
of interest. What is the effect of a positive money shock (t > 0) on the
nominal rate? How does this result compare to the = 1 case discussed
in the text? Explain.
To answer this problem, make use of the results in Section 5.7.3 of the Appendix to Chapter 5. Solving the basic model leads to equation (5.85) for the
price level:
pt =

d2 mt + a(1 + d2 )Et1 pt + Et pt+1 d2 vt + ut (1 + d2 )t


(1 + a)(1 + d2 )

39

(68)

Given the assumed process for mt , guess that the equilibrium price level is given
by
pt = b0 + b1 mt1 + b2 ut + b3 t + b4 vt + b5 t
Based on this guess,
Et1 pt = b0 + b1 mt1
and
Et pt+1 = b0 + b1 ( + mt1 + t )
Substituting these into (68), we have that the following condition must hold for
all realizations of mt1 and the random disturbances:
(1 + a)(1 + d2 )pt

= d2 ( + mt1 + t ) + a(1 + d2 ) [b0 + b1 mt1 ]


+ [b0 + b1 ( + mt1 + t )] d2 vt + ut (1 + d2 )t
= d2 + a(1 + d2 )b0 + b0 + b1 + [d2 + a(1 + d2 )b1 + b1 ] mt1
+ut (1 + d2 )t d2 vt + (d2 + b1 ) t

Using the proposed solution for pt , this requires that


(1 + a)(1 + d2 )b0 = d2 + a(1 + d2 )b0 + b0 + b1
(1 + a)(1 + d2 )b1 = d2 + a(1 + d2 )b1 + b1
(1 + a)(1 + d2 )b2 = 1
(1 + a)(1 + d2 )b3 = (1 + d2 )
(1 + a)(1 + d2 )b4 = d2
(1 + a)(1 + d2 )b5 = d2 + b1
Solving these yields the following solution for pt :
pt = b0 + b1 mt1 + b2 ut + b3 t + b4 vt + b5 t

pt



(1 + d2 )
d2
=
+
mt1
1 + d2
1 + d2


d2
ut (1 + d2 )t d2 vt
+
+
t
(1 + a)(1 + d2 )
(1 + a)(1 + d2 )
40

Using this result for pt , the equilibrium expressions for output can be obtained
from (5.81) as




d2
ut (1 + d2 )t d2 vt
+
yt = a
t + t
(1 + a)(1 + d2 )
(1 + a)(1 + d2 )





d2
ut d2 vt
1
+
= a
t +
t
(1 + a)(1 + d2 )
(1 + a)(1 + d2 )
1+a
while from (5.83) and (5.84),
it

= d2 (yt mt + pt + vt )




d2
d2
1
=
+
(ut + vt )
d2 (mt1 + t )
1 + d2
1 + d2
1 + d2

From these results, we can see that a positive realization of increases the
current price level and output and lowers the nominal rate of interest. Because
output rises, the real rate of interest rt must fall to ensure aggregate demand
and supply are equal at the temporarily higher level of output. When the money
supply follows a random walk ( = 1), a money supply shock has no effect on
the nominal interest rate, leading to a fall in the real rate but an equal rise in
expected inflation. When < 1, expected inflation rises less since the money
stock, after an initial increase, regresses back to its initial value.
To focus on the effect of on expected inflation, set u, v, and equal to
zero, and use the equilibrium expression for pt to obtain




d2
d2
+
( 1) mt1
Et pt+1 pt =
1 + d2
1 + d2


(1 + a) 1
+
d2 t
(1 + a)(1 + d2 )
while the impact on the real rate of interest is


a
rt = yt =
d2 t
(1 + a)(1 + d2 )
A rise in implies that has a larger impact on the expected future price level
since it has a larger impact on the future money supply. This means the current
price level rises more in response to a positive realizations of . In turn, the
larger unanticipated rise in pt generates a larger output increase and a larger
corresponding fall in the real rate of interest. Since pt rises more when is
large, expected future inflation is less affected since the rise in the level of the
money supply and the price level is more persistent.
10. An increase in average inflation lowers the real demand for money. Demonstrate this by using the model given by equations (5.34) - (5.37) and assuming the nominal money supply grows at a constant trend rate so
that mt = t to show that real money balance mt pt are decreasing in
.

41

Answering this problem involves repeating the steps outlined in Section 5.7.3
of the Chapter 5 Appendix, replacing the money supply process given in (5.86)
with the one in the problem. Since the focus is on the effects of the deterministic
trend on real money balances, it will simplify the problem if all stochastic
disturbances terms are ignored. In this case, the expression for the equilibrium
price level given in equation (5.85) becomes
pt

=
=

d2 mt + a(1 + d2 )Et1 pt + Et pt+1


(1 + a)(1 + d2 )
d2 t + a(1 + d2 )Et1 pt + Et pt+1
(1 + a)(1 + d2 )

(69)

For our guess for the solution, equation (5.87) is replaced by


pt = p0 + p t
so that Et1 pt = p0 + p t = pt and Et pt+1 = p0 + p (t + 1). Substituting these
into (69),

 

d2 t + a(1 + d2 ) p0 + p t + p0 + p (t + 1)
pt =
(1 + a)(1 + d2 )

d2 + a(1 + d2 )p + p
[1 + a(1 + d2 )] p0 + p
+
=
t
(1 + a)(1 + d2 )
(1 + a)(1 + d2 )
which equals p0 + p t if and only if

d2 + a(1 + d2 )p + p
p =
p =
(1 + a)(1 + d2 )
and
p0 =

[1 + a(1 + d2 )] p0 + p

=
(1 + a)(1 + d2 )
d2

With this expression for the equilibrium price level, real money balances will
equal



+ t =
mt pt = t
d2
d2
which is decreasing in the growth rate of nominal money balances . From the
solution for the price level, the rate of inflation is equal to . Higher values
of , and therefore higher rates of inflation, increase the opportunity cost of
holding money. This reduces the real demand for money, and, in equilibrium,
real money balances are lower at higher rates of money growth.

42

Chapter 6: Money and the Open Economy


1. Suppose mt = m0 + mt1 and mt = m0 + mt1 . Use equation (6.24)
to show how the behavior of the nominal exchange rate under flexible
prices depends on the degree of serial correlation exhibited by the home
and foreign money supplies.

Equation (6.24) on page 249 gives the following expression for the nominal
exchange rate:
st =

i


 

1

mt+i mt+i ct+i ct+i


1 + i=0 1 +

Using the result that ct+i ct+i = ct ct , this becomes


st = (ct

ct ) +

i



1

mt+i mt+i
1 + i=0 1 +

(70)

which is equation (6.25) of the text. We now have to use the specified processes

i 


for the nominal money supplies to evaluate the 1+


mt+i mt+i terms.
Since mt = m0 + mt1 ,
mt+1 = (m0 + mt ) = (1 + )m0 + 2 mt1
and
mt+2 = m0 (1 + + 2 ) + 3 mt1
Similarly,1
mt+i

= m0

j + i+1 mt1

j=0

= m0

1 i+1
+ i+1 mt1
1

For the foreign money supply,


mt+i = m0

1 ( )i+1
+ ( )i+1 mt1
1 ( )

1 This

uses the fact that 1 + a + a2 + ... + ak can be written as



 




1 + a + a2 + ... ak+1 + ak+2 + ...
=
1 + a + a2 + ... ak+1 1 + a + a2 + ...
=

for 1 < a < 1.

43

1
ak+1

1a
1a

Now, for notational easy, let b =

1+


. We need to evaluate

i
i=0 b



mt+i mt+i :




i+1
i+1



1

(
)
m0
=
bi mt+i mt+i
bi m0
1

1 ( )
i=0
i=0



bi i+1 mt1 ( )i+1 mt1

i=0

Taking each term individually,



1 i+1
i
b m0
=
1
i=0


m0 i 
b 1 i+1
1 i=0




m0
1
i i

b
1 1b
i=0



m0
1

1 1 b 1 b

=
=


i

( )i+1
1 ( )

1
m0

i=0

i i+1


=

m0
1

mt1 = mt1

i=0


i=0

bi ( )i+1 mt1 = mt1

i=0

st


i=0

1 b 1 b

bi i =

mt1
1 b

bi ( )i =

mt1
1 b

Now collecting all these results, equation (70) becomes








1
1
m0
1
m0

= (ct ct ) +

1 + 1 1 b 1 b
1 1 b 1 b


mt1 mt1

+
1 b
1 b

Since b = /(1 ), 1/(1 b) is equal to 1 + , and an expression like 1/(1 b)


is equal to (1 + )/(1 + (1 )). So we can write the nominal exchange rate as




m0

m0

st = (ct ct ) +
1
1

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



mt1
mt1

+
1 + (1 ) 1 + (1 )




1+
1+

= (ct ct ) + m0
m0
1 + (1 )
1 + (1 )



mt1
mt1

+
1 + (1 ) 1 + (1 )
44

To gain some insights from this expression, suppose that the money supplies
in both countries display the same autoregressive coefficient; = and m0 =
m0 . In this case,





mt1 mt1
st = (ct ct ) +
1 + (1 )
The nominal exchange rate depends on the initial difference in the money supplies. If = 1, this difference is permanent
 (the money supplies follow random
walks with drift), and st = (ct ct ) + mt1 mt1 , reflecting the permanent difference associated with the difference in price levels when mt1 = mt1 .
If < 1 but positive, then both mt and mt follow stable processes that converge
to m0 /(1 ). Any difference mt1 mt1 is now transitory and so has a
smaller impact on the current nominal exchange rate.
When = , the comparison is more complicated, since the money supplies
in the two countries regress towards their steady-state values at different rates.
The nominal exchange rate depends on the discounted value of the differences
in the paths followed by m and m .
2. In the model of Section 6.3 used to study policy coordination, aggregatedemand shocks were set equal to zero in order to focus on a common
aggregate-supply shock. Suppose instead that the aggregate supply shocks
are zero, and the demand shocks are given by u x + and u x +
so that x represents a common demand shock and and are uncorrelated country-specific demand shocks. Derive policy outcomes under
coordinated and (Nash) noncoordinated policy setting. Is there a role for
policy coordination in the face of demand shocks? Explain.
Problems 2 - 5 use the model of section 6.3, so it will be convenient to develop
some general expressions for output and inflation first, and then apply them to
the special cases considered in each of the problems. The basic model is given
in equations (6.35) - (6.39) on page 261 of the text. Equilibrium expressions
for output in the two countries are given by equations (6.43) and (6.44) on page
264.
Under a coordinated policy, the objective is to minimize

1  2
E yt + 2t + (yt )2 + ( )2
2
and the first order conditions for and are (see page 265),
b2 A1 yt + t + b2 A2 yt = 0

(71)

b2 A1 yt + t + b2 A2 yt = 0

(72)

45

It will prove convenient to use these to get expressions for t t , the difference
in inflation between the two countries, and t + t , the sum of inflation in the
two countries. First subtracting and then adding (71) and (72),
t t = b2 (A1 A2 )(yt yt )
t + t = b2 (yt + yt )
where we have used the fact that A1 + A2 = 1 (see page 264). Since the term
A1 A2 will appear frequently, let H A1 A2 .
Now using (6.43) and (6.44) to find yt yt and yt + yt , and substituting
the results into the expressions for t t and t + t , we obtain
t t

= b2 H [b2 H(t t )
+H(et et ) + 2A3 (ut ut )]


b2 H 2 (et et ) + 2HA3 (ut ut )
=
1 + b22 H 2

t + t

= b2 [b2 (t + t ) + et + et ]
b2 (et + et )
=
1 + b22

(73)

(74)

For later reference, we can also derive


yt yt

= b2 H( t t ) + H(et et ) + 2A3 (ut ut )





b2 H 2 (et et ) + 2HA3 (ut ut )
= b2 H
1 + b22 H 2
+H(et et ) + 2A3 (ut ut )


1
=
[H(et et ) + 2A3 (ut ut )]
1 + b22 H 2

and
yt + yt = b2

1
b2 (et + et )

+
e
=
+
e
(et + et )
t
t
1 + b22
1 + b22

(75)

(76)

Note from these expressions that average inflation in the two countries responds only to region-wide supply shocks (see equation 74), while inflation will
respond differently in the two countries to the extent that there are different
supply shocks or different demand shocks (see equation 73).
Adding together (73) and (74) yields, in the cooperative equilibrium,
2c,t =

b2 H [H(et et ) + 2A3 (ut ut )] b2 (et + et )

1 + b22 H 2
1 + b22
46

or2
c,t

1
et + et
H (et et )
b2 HA3 (ut ut )
= b2
+

2
2
2
2
1 + b2 H
1 + b2
1 + b22 (H)2



1 + b22 H 2 et 2A1 A2 (et et ) HA3 (ut ut )
+
= b2
(77)
[1 + b22 H 2 ] [1 + b22 ]
1 + b22 H 2

Then, from (74),


c,t

et + et
= c,t +
1 + b22



1 + b22 H 2 et + 2A2 A2 (et et ) HA3 (ut ut )

= b2
(78)
(1 + b22 H 2 ) (1 + b22 )
1 + b22 H 2

From (75) and (76),


 

1
1
[H(et et ) + 2A3 (ut ut )]
yc,t =
2
1 + b22 H 2
 

1
1
+
(et + et )
2
1 + b22
 


  

A1 1 + b22 H 2 et + A2 1 b22 H 2 et
A3 (ut ut )
=
(79)
+
(1 + b22 H 2 ) (1 + b22 )
1 + b22 H 2
and

yc,t




  

A1 1 + b22 H 2 et + A2 1 b22 H 2 et
A3 (ut ut )
=

(1 + b22 H 2 ) (1 + b22 )
1 + b22 H 2

(80)

These results all pertain to the case of a coordinated policy in the two countries. Under policy without coordination, each country takes the inflation rate
in the other as given in a Nash
The home countrys policy maker

 equilibrium.
sets inflation to minimize E yt2 + 2t while the foreign country policy maker
sets inflation to minimize E (yt )2 + ( )2 . The first order conditions take
the form
b2 A1 yt + t = 0
and
b2 A1 yt + t = 0
It follows that
t t = b2 A1 (yt yt )
2 This uses the fact that (A A )2 = (1 A A )2 = (1 2A )2 . Expanding the square
1
2
2
2
2
yields 1 4A2 (1 A2 ) = 1 4A1 A2 .

47

and
t + t = b2 A1 (yt + yt )
Using equations (6.43) and (6.44), these become
t t

= b2 A1 [b2 H(t t ) + H(et et ) + 2A3 (ut ut )]


b2 A1 [H(et et ) + 2A3 (ut ut )]
=
1 + b22 A1 H

(81)

and
t + t

= b2 A1 [b2 (t + t ) + et + et ]


b2 A1
=
(et + et )
1 + b22 A1

(82)

Adding these together,





b2 A1
b2 A1 [H(et et ) + 2A3 (ut ut )]

+
2 t =
(et + et )
1 + b22 A1 H
1 + b22 A1




2A1 1 + b22 H et + 2A2 et
2b2 A1 A3 (ut ut )
= b2 A1

[1 + b22 A1 H] [1 + b22 A1 ]
1 + b22 A1 H
or


N,t = b2 A1





A1 1 + b22 H et + A2 et
A3 (ut ut )
+
(1 + b22 A1 H) (1 + b22 A1 )
1 + b22 A1 H

(83)

From (82),
N,t



b2 A1
= N,t
(et + et )
1 + b22 A1



A1 1 + b22 H et + A2 et
= b2 A1
(1 + b22 A1 H) (1 + b22 A1 )



A3 (ut ut )

1 + b22 A1 H

(84)

The sum and differences of output in the noncooperative equilibrium are


yt yt =

H(et et ) + 2A3 (ut ut )


1 + b22 A1 H

and
yt +

yt


=

1
1 + b22 A1
48

(et + et )

(85)

(86)

Hence,



A1 1 + b22 H et + A2 et
A3 (ut ut )
+
=
(1 + b22 A1 H) (1 + b22 A1 )
1 + b22 A1 H


yN,t
and

yN,t




A1 1 + b22 H et + A2 et
A3 (ut ut )

(1 + b22 A1 H) (1 + b22 A1 )
1 + b22 A1 H

(87)

(88)

We can now address the specific questions posed in Problem 2. For this
problem, u x+ and u x+ so that x represents a common demand shock
and and are uncorrelated country-specific demand shocks, and et et 0.
thus, under a cooperative policy, equations (77), (78), (79), and (80) become


b2 HA3
c,t =
(t t )
1 + b22 H 2
c,t =


yc,t =
and

yc,t

b2 A3
1 + b22 H 2
A3
1 + b22 H 2

A3
=
1 + b22 H 2

(t t )

(t t )

(t t )

From (83), (84), (87), and (88), equilibrium inflation rates and outputs
without cooperation are


b2 A1 A3
N,t =
(t t )
1 + b22 A1 H
N,t =


yN,t =
and

=
yN,t

b2 A1 A3
1 + b22 A1 H
A3
1 + b22 A1 H

A3
1 + b22 A1 H
49

(t t )

(t t )

(t t )

Note that inflation response less to the relative demand shocks t t under
the coordinated policy than under the noncooperative policy as can be seen, for
example, by comparing the coefficients in the equilibrium expressions for c,t
and N,t :
b2 A1 A3
b2 HA3
<
2
2
1 + b2 H
1 + b22 A1 H
(To see that the inequality follows, rewrite the comparison as




b2 HA3 1 + b22 A1 H < b2 A1 A3 1 + b22 H 2
Dividing both sides by b2 A3 ,




H 1 + b22 A1 H < A1 1 + b22 H 2
which becomes H < A1 ; recalling that H = A1 A2 , and both A1 and A2 are
positive, this inequality always holds.) This contrasts with the case considered
in the text; with only a common supply shock, inflation responds more under a
coordinated policy (see page 267).
To investigate the potential role for policy coordination, we can evaluate the
home countrys loss function under the two policies. With coordination,
 
2 
2 

 2
1
b2 HA3
A3
+
+ 2

Lc =
2
2
2
2
1 + b2 H
1 + b2 H


2
 2

1
A3
+ 2
(89)
=
2
2 1 + b2 H
In the noncooperative equilibrium,
 
2 
2 

 2
1
b2 A1 A3
A3
+
+ 2
LN =

2
2
2
1 + b2 A1 H
1 + b2 A1 H





1 A23 1 + b22 A21  2
+ 2
=
2 [1 + b22 A1 H]2
Comparing (89) and (90), coordination yields a gain if and only if
LN > Lc
which occurs when
1 + b22 A21
[1 +

2
b22 A1 H]

>

1
1 + b22 H

This comparison reduces to


 

2

1 + b22 H 1 + b22 A21 > 1 + b22 A1 H
50

(90)

Multiplying out both sides, this becomes


1 + b22 H + b22 A21 + 2 b42 HA21 > 1 + 2b22 A1 H + 2 b42 A21 H 2
or
H(1 2A1 ) + A21 + b22 H(1 H)A21 > 0

(91)

But 1 2A1 = (A1 + A2 ) 2A1 = (A1 A2 ) = H, so (91) becomes


H 2 + A21 + b22 H(1 H)A21 > 0
Since A21 H 2 = (A1 H)(A1 + H) > 0 and 1 H > 0, the inequality holds.
Consider what happens in the face of a positive demand shock to the home
country ( > 0). The home country will deflate ( < 0) to partially offset
the impact of the demand shock on domestic output. Because the home policy
authority takes foreign inflation as given in a Nash equilibrium, it expects this
deflation to produce a real appreciation (see equation 6.42; falls when falls),
reducing the impact of inflation on domestic output. More inflation volatility is
needed to maintain output stability. Thus, offseting demand shocks is perceived
to be more costly. With a coordinated policy, is reduced while is increased,
thus serving to stabilize output will smaller fluctuations in inflation.
3. Continuing with the same model as in the previous question, how are real
interest rates affected by a common aggregate-demand shock?
In the notation of Problem 2, x was a common aggregate demand shock.
As was shown as part of the solution to Problem 2, neither home nor foreign
inflation is affected by a common demand shock (equations 77, 78, 79, 80, 83,
84, 87, and 88 all depended only on u u from which a common demand shock
cancels out). With output and inflation independent of x, it must be that a
common demand shock alters real interest rates to offset the demand shock and
maintain aggregate demand constant (since output remains constant). Thus, a
positive value of x should raise real interest rates in both countries; a negative
x should lower real rates.
To verify this, first note from equation (6.42) on page 263 that a common
demand shock will leave the real exchange rate unchanged; the right side of (6.42)
depends only on terms like , e e , and u u , none of which are affected
by a common demand shock (see equations (73) and 81) With unaffected, the
interest parity condition (6.39) implies that r r must remain unchanged, so
both interest rates change by the same amount. Adding together the aggregate
demand equations (6.37) and (6.38), we can obtain
 
1
rt + rt =
[ut + ut (1 a3 )(yt + yt )]
a2

51

If the x disturbance is the only shock, ut + ut = 2x and, from either (76) or


(86), yt + yt is unaffected, so
 
1

rt + rt = 2
x
a2
Since we have seen that with only an x shock, r r = 0, it follows that
 
1

rt = rt =
x
a2

4. Policy coordination with asymmetric supply shocks: Continuing with the


same model as in the previous two questions, assume there are no demand
shocks but that the supply shocks e and e are uncorrelated. Derive policy
outcomes under coordinated and uncoordinated policy setting. Does coordination or noncoordination lead to greater or smaller inflation response
to supply shocks? Explain.
From equations (77), (78), (79), (80), (83), (84), (87), and (88) that were
derived in the process of solving Problem 2, the outcomes under coordinated and
noncoordinated policies when u u 0 are

c,t

c,t




1 + b22 H 2 et 2A1 A2 (et et )
= b2
[1 + b22 H 2 ] [1 + b22 ]



1 + b22 H 2 et + 2A1 A2 (et et )
= b2
[1 + b22 H 2 ] [1 + b22 ]


yc,t




 
A1 1 + b22 H 2 et + A2 1 b22 H 2 et
=
(1 + b22 H 2 ) (1 + b22 )


yc,t




 
A1 1 + b22 H 2 et + A2 1 b22 H 2 et
=
(1 + b22 H 2 ) (1 + b22 )





A1 1 + b22 H et + A2 et
(1 + b22 A1 H) (1 + b22 A1 )




A1 1 + b22 H et + A2 et
(1 + b22 A1 H) (1 + b22 A1 )

N,t = b2 A1

N,t = b2 A1

52

yN,t
and




A1 1 + b22 H et + A2 et
=
(1 + b22 A1 H) (1 + b22 A1 )


yN,t




A1 1 + b22 H et + A2 et
=
(1 + b22 A1 H) (1 + b22 A1 )

Under a coordinated policy, the response to a domestic supply shock home


inflation is

1 + b22 H 2 2A1 A2
et
c,t = b2
(1 + b22 H 2 ) (1 + b22 )
while under noncoordinated policy it is




A21 1 + b22 H
N,t = b2
et
(1 + b22 A1 H) (1 + b22 A1 )
These expressions are messy, so an easier way of comparing outcomes is to
return to equation (73) and (74) for the case of cooperation and equations (81)
and (82) for the noncooperation case. For the sum of inflation rates, (74) and
(82) imply that t +t responds more to et under cooperation that in the absence
of cooperation (the coefficient on e in 82 is increasing in A1 which is less than
1 and under cooperation, the response in 74 is obtained by setting A1 = 1).
So ( t + t )c responds more than (t + t )N . On the other hand, ( t t )c
responds less than (t t )N as can be seen by comparing (73) and (81).3
The results that (t + t )c > ( t + t )N and ( t t )c < (t t )N allows
us to conclude that foreign inflation responds more (add the two inequalities
together), but it does not resolve whether domestic inflation responds more. We
saw in the text that cooperation leads to a larger inflation response to a common
supply shock than occurred without cooperation. Under cooperation in the face
of a home country supply shock, more of the adjustment is made by foreign
inflation than would occur without cooperation, and this acts to allow domestic
inflation to respond less. So inflation rate in the two countries diverge less
( ( t t )c < (t t )N ). But cooperation tends to lead to a stronger overall
response ( ( t + t )c > (t + t )N ), so the net effect on t is not clear.
5. Assume the home-country policy maker acts as a Stackelberg leader and
recognizes that foreign inflation will be given by equation (6.47). How
does this change in the nature of the strategic interaction affect the home
countrys response to disturbances?
3 This uses the fact that A H > H 2 since A > A (see the definitions of the A s on page
1
1
2
i
264).

53

Referring to section 6.3.3 of the text, the home policy authority picks inflation to minimize yt2 + 2t , but now, rather than taking foreign inflation t
as given, the home policy authority recognizes that t will be set according to
equation (6.47). Equation (6.47) was derived for the case of a single common
supply shock; et = et = t . The first order condition for the home country will
reflect this dependence of t on t . Thus, the first order condition for the home
countrys choice of inflation will be

 2


b2 A1 A2
b2 A1 b2 A2
(92)
yt + t = 0
1 + b22 A21
This should be compared to the expression
above equation (6.46) on
 2 immediately

b2 A1 A2
page 266. the extra term b2 A2 1+b2 A2 yt is the effect of that the home
2 1
countrys inflation rate has on home country output by causing foreign inflation
to adjust based on the reaction function given by (6.47).
To solve (92), first note that the coefficient on yt can be rewritten as
 2


b2 A1 A2
1 + b22 H
b2 A1 b2 A2
= b2 A1
1 + b22 A21
1 + b22 A21
since (A21 A22 ) = H(A1 + A2 ) and A1 + A2 = 1. If we define
S

1 + b22 H
<1
1 + b22 A21

then the first order condition for the home country becomes
b2 A1 Syt + t = 0
which contrasts with the first order conditon in the Nash case ( b2 Ayt +t = 0).
One why to interpret this is that because S < 1, the marginal output efects of a
rise in home inflation are now smaller, since the home policy maker recognizes
that higher will induce the foreign country to reduce its inflation rate, causing
a depreciation for the home country ( rises; see 6.42) that acts to offset the
expansionary impact of the rise in domestic inflation (see 6.35). It will be
optimal for to response less. .
Also, using the foreign countrys reaction function, home country output
(from 6.43) can be written

 2



b2 A1
b2 A1 A2
t +
t + t
yt = b2 A1 t b2 A2
1 + b22 A21
1 + b22 A21

1 + b22 A1 H
1 + b22 H
= b2 A1
+

t
t
1 + b22 A21
1 + b22 A21
Now use the expression for home country output (equation 6.43 of the text) to
write the first order condition (92) as


1 + b22 A1 H
1 + b22 A1 H
1 + b22 H
A
+
b

t + t = 0
b2 A1
2
1
t
1 + b22 A21
1 + b22 A21
1 + b22 A21

54

or


t =


2
b2 A1 1 + b22 A1 H

(1 + b22 A21 S s ) (1 + b22 A21 )

where
S
s


t




1 + b22 A1 H 1 + b22 H
2

(1 + b22 A21 )

From (83), the response to a common supply shock in a Nash equilibrium is


b2 A1
t =
t
(1 + b22 A1 )

6. In a small open economy with perfectly flexible nominal wages, the text
showed that the real exchange rate and domestic CPI were given by


+ et+i ut+i
a2 rt+i
di Et
t =
a1 + a2 + b1
i=0
and

1
pt =
1 + c i=0

1
1+c

i

Et [mt+i zt+i vt+i ]

where zt+i yt+i + (1 h)t+i crt+i . Assume r = 0 for all t and


that e, u, and z + v all follow first order autoregressive processes (e.g.
et = e et1 + xet for xe white noise). Let the nominal money supply be
given by
mt = g1 et1 + g2 ut1 + g3 (zt1 + vt1 )
Find equilibrium expressions for the real exchange rate, the nominal exchange rate, and the consumer price index. What values of the parameters
g1 , g2 , and g3 minimize fluctuations in st ? in qt ? in t ? Are there any
conflicts between stabilizing the exchange rate (real or nominal) and stabilizing the consumer price index?
The first thing to note is that under the assumptions of the problem, the real
exchange rate t is independent of the money supply process, depending only on
the exogenous behavior of r , et , and ut . Thus, in this example, the behavior of
t is unaffected by the choice of the gi parameters. We can use the assumptions
of the problem to write the real exchange rate as
i




i

1
1
1
i e et u ut
d
et
ut
=
(93)
t =
a1 + a2 + b1
a1 + a2 + b1
1 de
1 du
i=0
55

To evaluate the expression for the equilibrium price pt , we do need to use


the money supply process:

pt

1
1 + c i=0

1
1 + c i=0




1
1+c
1
1+c

i
i

Et [g1 et1+i + g2 ut1+i + g3 t1+i t+i ]


 i

g1 e et1 + g2 iu ut1 + g3 i t1 i ( t1 + xt )

where t zt vt . Collecting terms,


pt

i

 i

1
1
g1 e et1 + g2 iu ut1 + (g3 ) i t1 i xt
=
1 + c i=0 1 + c

 

g1 et1
g2 ut1
=
+
1 + c(1 e )
1 + c(1 u )
 


xt
(g3 ) t1

+
(94)
1 + c(1
1 + c(1 )

So pt is stabilized if g1 = g2 = 0 and g3 = . Since neither e nor u affect p in


this setup, any response by m to these disturbances would simply add additional
variance to the price level. By setting g3 = , policy is able to insulate pt from
the forecastable movements in t .
The nominal exchange rate st is given by st = t p + pt where p is the
foreign price level. For simplicity, set p = 0. Then st = t + pt . Combining
(93) and (94), the nominal exchange rate is



1
1
1
(e et1 + xet )
(u ut1 + xut )
st =
a1 + a2 + b1
1 d
1 du

 
 e
 

g1 et1
xt
g2 ut1
(g3 ) t1
+

+
+
1 + c(1 e )
1 + c(1 u )
1 + c(1
1 + c(1 )
The effects of shocks on the nominal exchange rate are minimized if g3 = .
The impact of et1 is eliminated if


 

1
e
g1
+
=0
a1 + a2 + b1
1 de
1 + c(1 e )
or

1 + c(1 e )
g1 =
a1 + a2 + b1



e
1 de

The effects ut1 on st can be eliminated if




 

1
u
g2

+
=0
a1 + a2 + b1
1 du
1 + c(1 u )

56

or


g2 =

1 + c(1 u )
a1 + a2 + b1



u
1 du

In terms of consumer prices qt , from equation (6.52) on page 270,


qt

= hpt + (1 h)(st + pt )
= pt + (1 h)t

(95)

Combining this with (93) and (94),



 
 
 

g1 et1
xt
g2 ut1
(g3 ) t1
qt =

+
+
1 + c(1 e )
1 + c(1 u )
1 + c(1
1 + c(1 )



1
1
1h
(e et1 + xet )
(u ut1 + xut )
+
a1 + a2 + b1
1 de
1 du
The effects of shocks on the consumer price index are minimized if g3 = .
The impact of et1 is eliminated if



 
g1
e
1h
=0
+
1 + c(1 e )
a1 + a2 + b1
1 de
or


g1 = (1 h)

1 + c(1 e )
a1 + a2 + b1



e
1 de

The effects ut1 on qt can be eliminated if





 
g2
u
1h
=0

1 + c(1 u )
a1 + a2 + b1
1 du
or


g2 = (1 h)

1 + c(1 u )
a1 + a2 + b1



u
1 du

There are no conflicts (at least in this example) between stabilizing the nominal exchange rate and stabilizing the consumer price level in the face of shocks
( z v). Since has no effect on the real exchange rate, stabilizing domestic
prices would also stabilize the nominal exchange rate and the consumer price
level. For e and u disturbances, the appropriate responses of m to stabilize q
are proportional to the optimal responses to stabilize s, but the responses are
smaller in absolute value by a factor 1 h if the objective is to stabilize q. Both
e and u affect the real exchange rate. Since qt = pt + (1 h)t , policy can
stabilize q by making p move to offset any movement in (1 h). The nominal
exchange rate, however, is equal to t + pt , so it is stabilized it p fully adjusts
to offset movements in .

57

7. Equation (6.42) for the equilibrium real exchange rate in the two-country
model of section 6.3.1 takes the form t = AEt t+1 + vt . Suppose vt =
vt1 + t , where t is a mean-zero, white-noise process. Suppose the
solution for t is of the form t = bvt . Find the value of b. How does it
depend on ?
First note that we can think of v as having a direct impact on , holding
the expected future real exchange rate constant, and an indirect effect if v alters
Et t+1 . Under the proposed solution, Et t+1 = bEt vt+1 = bvt . Substituting
this into the equilibrium condition for the real exchange rate,
t = Abvt + vt = (1 + Ab) vt
This can equal bvt for all realizations of vt if only if
b = (1 + Ab)
from which it follows that
b=

1
1 A

The expression for b shows that a rise in increases b and leads an innovation in v to have a larger impact on the real exchange rate. This can be seen
more clearly by writing the solution for the real exchange rate as
t = bvt1 + bt
If is large (i.e., close to 1 say), then innovations to the v process are very
persistent. Therefore, in addition to the direct impact of v on , there will be a
larger impact on Et t+1 if is large. Thus, the total impact of an innovation
t on the current real exchange rate t will be larger (i.e., b is larger).

58

Chapter 8: Discretionary Policy and Time Inconsistency


1. Assume firms maximize profits in competitive factor markets with labor
the only variable factor of production. Output is produced according to
the production function Y = AL , 0 < < 1. Labor is supplied inelasticly. Nominal wages are set at the start of the period at a level consistent
with market clearing, given expectations of the price level. Actual employment is determined by firms once the actual price level is observed. Show

1
(p pe ) + 1
ln A,
that, in log terms, output is given by y = l + 1

where l is the log labor supply. (Note: the text contains a typo: al
appears instead of l .)

Given the assumed form of the production function, the demand for labor
can be obtained from the condition that firms set the marginal product of labor,
AL1 , equal to the real wage W/P :

Ld =

W
AP

1
 1

If L is the fixed supply of labor, the equilibrium real wage that equates labor
demand and labor supply is
 
W
= A (L )1
P
or, in log terms,
(w p) = ln A (1 )l
If workers and firms set the nominal wage at the start of the period, prior to
observing actual prices, then the contract wage consistent with labor market
clearing is
wc = Ep + ln A (1 )l
Actual (log) employment, given by the demand for labor, will be
1
(wc p ln A)
1

1 
Ep p (1 )l
=
1
1

(p Ep)
= l +
1

l =

and the log of output is equal to


y

= ln A + l
= ln A +

(p Ep) + l
1
59

2. Suppose an economy is characterized by the following three equations:


= e + ay + e
y = br + u
m = di + y + v
where the first equation is an aggregate supply function written in the
form of an expectations-augmented Phillips Curve, the second is an IS or
aggregate-demand relationship, and the third is a money demand equation
where m denotes the growth rate of the nominal money supply. The real
interest rate is denoted by r and the nominal rate by i, with i = r + e .
Let the monetary authority
implement

 policy by setting i to minimize the
expected value of 12 (y y )2 + 2 where y > 0. Assume the policy
authority has forecasts ef , uf , and v f of the shocks, but the public forms
its expectations prior to the setting of i and without any information on
the shocks.
(a) Assume the monetary authority can commit to a policy of the form
i = c0 + c1 ef + c2 uf + c3 vf prior to knowing any of the realizations of
the shocks. Derive the optimal commitment policy (i.e., the optimal
values of c0 , c1 , c2 , and c3 ).
(b) Derive the time-consistent equilibrium under discretion. How does
the nominal interest rate compare to the case under commitment?
What is the average inflation rate?
(a) From the IS relationship, y = b(i e ) + u, so if we now use the
aggregate supply relationship, = e ab(i e ) + au + e, or
= (1 + ab)e abi + au + e

(96)

Taking expectations of both sides, conditional on the publics information, e =


ie . Hence, inflation is
= (1 + ab)ie abi + au + e
and the objective function, expressed in terms of the policy instrument i becomes

1 
(97)
L E (b(i ie ) + u y )2 + [(1 + ab)ie abi + au + e]2
2
Under the commitment policy, the central bank follows a policy of the form
i = c0 + c1 ef + c2 uf + c3 vf
60

(98)

where xf denotes the central banks forecast of x. With this policy rule, ie = c0 .
Substituting this result and the policy rule into (97) gives
L


2
1
E b(c1 ef + c2 uf + c3 v f ) + u y
2

2

1 
+ E c0 ab c1 ef + c2 uf + c3 vf + au + e
2

(99)

The objective is to minimizing this function by the choice of c0 , c1 , c2 and


c3 where the choices are made prior to actually observing any of the shocks or
forecasts. For c0 , the first order condition is




E c0 ab c1 ef + c2 uf + c3 vf + au + e = c0 = 0
For c1 , the first order condition (using c0 = 0) is

 

0 = E b(c1 ef + c2 uf + c3 vf ) + u y (bef )




+E ab c1 ef + c2 uf + c3 vf + au + e (abef )
If the shocks are mutually uncorrelated, this becomes
c1 ( + a2 )b2 fe abfe = 0
where fe is the variance of the forecast of e and the result from rational expectations that E(eef ) = fe has been used Hence,
c1 =

a
( + a2 )b

For c2 , one obtains



 

0 = E b(c1 ef + c2 uf + c3 vf ) + u y (buf )




+E ab c1 ef + c2 uf + c3 vf + au + e (abuf )
or
c2 ( + a2 )b2 fu ( + a2 )bfu = 0
yielding
c2 =

1
b

The first order condition for c3 is




0 = E b(c1 ef + c2 uf + c3 v f ) + u y (bvf )




+E ab c1 ef + c2 uf + c3 vf + au + e (abvf ))
or
c3 = 0
61

Hence, the optimal commitment policy is




1
a
f
e
ic =
uf +
b
+ a2
which does not depend on the money demand function at all.
(b) Under discretion, the central bank treats expectations as given in choosing
i to minimize its expected loss function, based on its forecasts of the underlying
shocks. In this case, the expectations operator in the loss function (97) is conditional on ef , uf , and vf . The first order condition for the choice of i under
discretion is




(b) b(i ie ) + uf y ab (1 + ab)ie abi + auf + ef = 0
or




b2 (i ie ) buf + by ab (1 + ab)ie abi + auf + ef = 0
Solving for i,
i=

(b + a(1 + ab)) ie y + ( + a2 )uf + aef


b( + a2 )

(100)

Taking expectations based on the publics information,


ie =

y
a

Hence, (100) becomes


y 1
i=
+
a
b


uf +

a
ef
+ a2


(101)

Notice that policy under discretion responds to the stochastic shocks the same
way as policy would if commitment were possible. However, the nominal rate
under discretion is systematically higher than under commitment. Since expected
inflation is equal to ie ,
e =

y
>0
a

3. Verify that the optimal commitment rule that minimizes the unconditional
a
expected value of the loss function given by (8.10) is mc = 1+a
2 e.

62

The loss function (8.10) is


1
1
2
2
V c = [a(b1 e + v) + e k] + [b0 + b1 e + v]
2
2
Under the commitment policy, the central bank chooses b0 and b1 to minimize
the unconditional expectation of V c . If the shocks are uncorrelated, we can write
this expectation as
 1 2

1 
b + b21 2e + 2v
EV c = (1 + ab1 )2 2e + a2 2v + k2 +
2
2 0
where 2x is the variance of x. If we minimize this with respect to b0 and b1 ,
the first order condition for b0 is
b0 = 0
while that for b1 is
a(1 + ab1 ) 2e + b1 2e = 0
or b1 = a/(1 + a2 ). Hence, the optimal commitment policy is
mc =

a
e
1 + a2

as claimed.
4. Suppose the central bank acts under discretion to minimize the expected
value of equation (8.2). The central bank can observe e prior to setting
m, but v is observed only after policy is set. Assume, however, that e
and v are correlated, and that the expected value of v, conditional on e,
is E [v|e] = qe where q = v,e /2e and v,e is the covariance between e and
v.
(a) Find the optimal policy under discretion. Explain how policy depends on q.
(b) What is the equilibrium rate of inflation? Does it depend on q?
(a) The loss function (8.2) is quadratic in inflation and the deviation of
output from a target level. Taking the model to consist of equations (8.2) (8.4), the central banks objective function can be written as
EV

=
=

1
1
E [a( e ) + e k]2 + E []2
2
2
1
1
e
E [a(m + v m ) + e k]2 + E [m + v]2
2
2

63

If the central bank takes private sector expectations as given and can observe e
prior to setting policy, then the first order condition for the optimal choice of
m is
a [a(m + E [v|e] me ) + e k] + m + E [v|e] = 0
or
a [a(m + qe me ) + e k] + m + qe = 0
Hence,


a2 me + ak 1 + q(1 + a2 e
m =
1 + a2
Taking expectations, conditional on the publics information set,
me = ak
so



1 + q(1 + a2 e
m = ak
1 + a2

(102)

The optimal respond to e will depend on q. If the central bank could observe
v, it would adjust m to offset the impact of v on inflation. If e provides
some information on which the central bank can base a forecast of v, then it
will adjust m to offset the forecasted impact of v on inflation. To see this,
note that (102) can be written as
1
e qe
1 + a2
1
e E [v|e]
= ak
1 + a2

m = ak

(1+q(1+a2 )e
(b) The equilibrium rate of inflation is m + v or ak
+ v.
1+a2
this depends on q since the central banks choice of money growth is a function
of q. The average rate of inflation, or the inflation bias, is, however, equal to
ak, and this is independent of q.
5. Since the tax distortions of inflation are related to expected inflation,
suppose the loss function (8.2) is replaced by
2

L = (y yn k)2 + (e )

where y = yn + a( e ). How is Figure 8.2 modified by this change in


the central banks loss function? Is there an equilibrium inflation rate?
Explain.

64

The inflation loss is now assoicated with expected inflation. Since the central
bank is assumed, under discretion, to take expected inflation as given, it will now
view the costs of inflation as given, so increasing inflation a bit is perceived by
the central bank to yield benefit in terms of higher output but no cost.
To see this, substitute the aggregate supply relationship into the loss function
to obtain
2
2
L = [a( e ) k] + ( e )

If this is minimized with respect to , the first order condition is


2a [a( e ) k] = 0
or
= e +

k
a2

(103)

There is no expected rate of inflation such that e = e + k/a2 . Figure 8.2,


giving the central banks optimal choice of inflation, as a function of the publics
expected rate of inflation, shows that the central banks reaction function given
by (103) does not cross the 45 line.

6. Based on Jonsson (1995) and Svensson (1997b). Suppose equation (8.3)


is modified to incorporate persistence in the output process:
yt = (1 )yn + yt1 + a( t et ) + et ;

0<<1

Suppose the policy maker has a two-period horizon with objective function
given by
L = min E [Lt + Lt+1 ]


where Li = 12 (yi yn k)2 + 2i .
(a) Derive the optimal commitment policy.
(b) Derive the optimal policy under discretion without commitment.
(c) How does the presence of persistence ( > 0) affect the inflation bias?
(a) Under commitment, the central bank follows a rule that specifies how
inflation will be set as a function of the state of the economy. The rule is chosen
before the central bank knows the current state of the economy, and the public
uses the rule to form their expectations about policy. They can do so since, by
assumption, the central bank is committed to following the rule. Since the state

65

Figure 1: Dashed line is the central banks reaction function.


0.35

Central Bank's Planned Inflation Rate

0.3

0.25

0.2

0.15

0.1

0.05

0
0

0.05

0.1

0.15

0.2

0.25

0.3

Expected Inflation

Figure 2:
at time t is characterized by yt1 and et , one can specify the commitment policy
as
t = b0 + b1 xt1 + b2 et

(104)

t+1 = b0 + b1 xt + b2 et+1

(105)

for period t, and

for period t + 1 where


xt1 yt1 yn
will be used to denote the output gap. Because the policy choice at time t may,
through xt , affect output at time t + 1, while any future effect of policy in t + 1
doesnt matter (since the loss function only involves Lt and Lt+1 ), the optimal
response to et may differ from the optimal response to et+1 . For this reason,
the coefficients in (104) and (105) are allowed to differ.
We need to find the optimal values of the coefficients in the policy rules that
minimize the unconditional value of the loss function L. Here, we can think of
the central bank deciding on the parameters of the policy rule prior to having
any information about the state of the economy. Thus, it evaluates things from
the perspective of the unconditional expectation of et (equal to zero) and xt1
(also equal to zero)
66

The public knows the value of xt1 when forming expectations of time t, so
using the policy rule (104),
et = b0 + b1 xt1
and
t et = b2 et
Using the aggregate supply relationship,
xt = xt1 + (1 + ab2 )et
Similarly, t+1 et+1 = b2 et+1 and
xt+1

= xt + (1 + ab2 )et+1
= 2 xt1 + (1 + ab2 )et + (1 + ab2 )et+1

Substituting these expressions into the loss function yields



1 
E [xt1 + (1 + ab2 )et k]2 + (b0 + b1 xt1 + b2 et )2
E [Lt + Lt+1 ] =
2

2
1
+ E 2 xt1 + (1 + ab2 )et + (1 + ab2 )et+1 k
2
1
2
+ E [b0 + b1 (xt1 + (1 + ab2 )et ) + b2 et+1 ]
2
Now minimize this with respect to the parameters in the policy rule. Doing so
yields the following first order conditions:
For b0 :
E [b0 + b1 xt1 + b2 et ] = 0
or
b0 = 0
For b1 :
E [(b0 + b1 xt1 + b2 et ) xt1 ] = 0
or
b1 = 0
For b2 :
0 = E [(xt1 + (1 + ab2 )et k) aet ] + (b0 + b1 xt1 + b2 et ) et



+E 2 xt1 + (1 + ab2 )et + (1 + ab2 )et+1 k aet
+E [(b0 + b1 (xt1 + (1 + ab2 )et ) + b2 et+1 ) b1 aet ]
67

or a(1 + ab2 )2e + b2 2e + 2 a(1 + ab2 )2e + a (b1 ) (1 + ab2 )2e = 0. Solving
for b2 ,
b2 =

a + a2 + a (b1 )2

1 + a2 + a2 2 + a2 (b1 )2

which depends on b1 . So turning to the first order condition for b1 ,
E [(b0 + b1 (xt1 + (1 + ab2 )et ) + b2 et+1 ) (xt1 + (1 + ab2 )et )] = 0
or b1 E (xt1 + (1 + ab2 )et )2 = 0 which implies
b1 = 0
Substituting this into the expression for b2 ,

 

a 1 + 2


b2 =
1 + a2 1 + 2
The first order condition for the optimal b2 is


 
0 = E 2 xt1 + (1 + ab2 )et + (1 + ab2 )et+1 k aet+1
+E [(b0 + b1 (xt1 + (1 + ab2 )et ) + b2 et+1 ) et+1 ]
or
a(1 + ab2 ) 2e + b2 2e = 0
which yields
b2


=

a
1 + a2

Finally, it is straightforward to show that b0 = 0. Thus, the optimal commitment


policy takes the form

 

a 1 + 2
c
 et

(106)
t =
1 + a2 1 + 2
and

ct+1

a
=
1 + a2


et+1

(107)

Notice that average inflation is zero in each period. Also, the response to
et+1 given by b2 is exactly the same as obtained in, for example, equation (8.11)
of the text. Since in period t + 1 we have a standard one period problem, this is
not surprising. In period t, however, the optimal response to et differs from the
68

standard optimal response to a supply shock in a one-period model if is nonzero. Assuming > 0 (the central bank cares about both period), the optimal
commitment policy in period t reduces to the standard result if = 0. If = 0,
then the optimal response to a period t supply shock as affected. Suppose > 0.
Then it becomes optimal to offset more of the impact of et on period t output
by responding more strongly to et ;


a 1 + 2
a
>

1 + a2
1 + a2 1 + 2
Since xt will affect xt+1 , period t + 1 output can be made more stable by insulating xt more from et .
(b) Under discretion, the central bank picks in each period, taking expectations and the previous periods output as given. Solving backwards, the central
bank will choose inflation in period t + 1 to minimize
2 1
1 
xt + a(t+1 et+1 ) + et+1 k + t+1
2
2
taking xt and et+1 as given. The first order condition is


a xt + a( t+1 et+1 ) + et+1 k + t+1 = 0
or
t+1 =

a2 et+1 a (xt + et+1 k)


1 + a2

Hence,
et+1 = ak axt
Expected inflation is decreasing in last periods output if output displays positive
persistence ( > 0). A positive value of xt means that output in period t + 1
will be closer to the desired level yn + k. Thus, the central banks incentive to
inflate is reduced. Anticipating this, the public expects lower inflation.
Actual inflation in period t + 1 under discretion will equal


a
d
(108)
t+1 = ak axt
et+1
1 + a2
and output will be

yt+1 = yn + (yt yn ) +

1
1 + a2


et+1

Notice that the response to et+1 is the same as under the optimal committment
policy given by equation (107).

69

The central banks loss in period t + 1 is


2

1
1
1
a
e
e

k
+

Lt+1 = xt +
ak

ax
t+1
t
t+1
2
1 + a2
2
1 + a2
Now use these results to evaluate the central banks decision problem in period
t. The two period loss function is

2
1
1
2
2
et+1 k
L = E [xt k] + Et + E xt +
2
1 + a2

2
1
a
e
+ E ak axt
t+1
2
1 + a2
Since xt = xt1 + a(t et ) + et , the first order condition for the optimal
inflation rate in period t, taking xt1 and expectations as given, is

1
et+1 k
0 = a [xt k] + t + aE xt +
1 + a2

a
et+1
a2 E ak axt
1 + a2
or
0 = a [xt1 + a(t et ) + et k] + t
+a [(xt1 + a(t et ) + et ) k]
a2 [ak a (xt1 + a(t et ) + et )]
Solving for t ,



 



2
a2 1 + 2 + a2 2
a
1
+

+
a


 et +

 k
t =
1 + a2 1 + 2 + a2 2
1 + a2 1 + 2 + a2 2




1 + 3 + a2 3
1 + 2 + a2 2

 xt1 a

 et
a
1 + a2 1 + 2 + a2 2
1 + a2 1 + 2 + a2 2
It follows that




et = a 1 + + a2 k a 1 + 3 + a2 3 xt1
and inflation under discretion is equal to




dt = a 1 + (1 + a2 ) k a 1 + 3 + a2 3 xt1


1 + 2 (1 + a2 )

 et
a
1 + a2 1 + 2 (1 + a2 )

70

(109)

(c) Optimal discretionary policy is characterized by equations (109) for period


t and (108) for period t+1. In period t, the fact that output displays persistence
does affect the inflation bias. Creating an output expansion in period t will tend
to raise output in period t + 1, so this increases the incentive to expand output
in period t. Anticipating this,
 higher inflation in period t, and
 the public expects
average inflation equals a 1 + (1 + a2 ) k which exceeds the one-period bias
ak. The inflation bias is increasing in since a larger implies that any output
expansion in t will lead to a larger expansion in t + 1. In addition, a low value
of yt1 (so xt1 is negative) acts to move yt further from the central banks
desired level yn k; this increases the incentive to inflate and raises the expected
and actual rate of inflation. Comparing (109) and (106) shows that the period t
response to et is also distorted under discretion. this is in contrast to the result
we have usually found. When = 0 , however, the response to et is larger, in
absolute value, under discretion
7. Show that d given by (8.18) is equal to inflation under discretion when the
weight on inflation in (8.2) becomes 1+ and the economy is characterized
by (8.3) and (8.4).
The central banks decision problem under discretion can be written as


1
1
2
e
2
[a( ) + e k] + (1 + )
min
2
2
where = m + v and the central bank observes e but not v before
determining policy. The first order condition is
a [a(m me ) + e k] + (1 + ) m = 0
or
m =

a2 me ae + ak
1 + + a2

It follows that me = ak/(1 + ). Hence




ak
ae + ak
a2 1+
m =
1 + + a2


ak
ae
=

1+
1 + + a2
and inflation is

=

ak
1+

as claimed in equation (8.18).

71

ae
+v
1 + + a2

8. Suppose that the private sector forms expectations according to


et = if t1 = et1
et = ak otherwise.
If the central banks objective function is given by (8.2) and its discount
rate is , what is the minimum value of that can be sustained in
equilibrium?
The central banks single period loss function, from (8.2), is
Lt =


1
(yt yn k)2 + 2t
2

Assume output is given by equation (8.3) of the text:


yt = yn + a( t et )
where the aggregate supply shock has been set to zero to parallel the analysis of
reputation in section 8.3.1.1. To determine the incentive the central bank has
to deviate from maintaining an inflation rate of , we need to consider what
happens if the central bank deviates in period t, incurs the punishment in period
t + 1 (the punishment being that the public expects an inflation rate of ak) and
then returns to an inflation rate of in period t + 2. Let Lnd
t be the loss from
no-deviation in period t and Ldt the loss from deviating. The central bank will
wish to deviate if
nd
Ldt + Ldt+1 < Lnd
t + Lt+1

Following the discussion in the text, we can say that the central bank will have
an incentive to deviate if the gain exceeds the cost, or

 d
d
nd

G( ) Lnd
(110)
t Lt > Lt+1 Lt+1 C( )
To evaluate these terms, start with the loss under the no-deviation case.
Not deviating means the central bank sets the rate of inflation equal to each
period. This is expected by the public, so output is equal to yn . The loss each
period is then

1 2
nd
k + ( )2
Lnd
t = Lt+1 =
2
If the central bank deviates in period t, then the public will expect an inflation
rate equal to the one-shot discretionary rate ak in period t + 1. The best the
central bank can do is to inflation at this rate, ensuring yt+1 = yn and
 1 

1 2
k + (ak)2 = 1 + a2 k2
Ldt+1 =
2
2
72

Now we can determine Ldt . Since the public is expecting , the central bank,
if it deviates, will pick t to minimize
Ldt =


1
(a(t ) k)2 + 2t
2

The first order condition is


(1 + a2 )t a2 ak = 0
or
t =

a2 + ak
1 + a2

Output in period t will equal


yt

and
Ldt

=
=
=

= yn + a(t )


ak
= yn + a
1 + a2

  

2  2
2 
1
a + ak
ak
a
k +
2
1 + a2
1 + a2
 
2  2
2 
1
a + ak
k + a
+

2
1 + a2
1 + a2
1 (k + a )2

2 1 + a2

We can now evaluate the gains and cost of deviation:


d
G( ) = Lnd
t Lt
 1 (k + a )2

1
k2 + ( )2
=
2
2 1 + a2

and


C( ) = Ldt+1 Lnd
t+1


 2 1 2
1 
2
2
1+a k
k + ( )
=
2
2


1
a2 2 k2 ( )2
=
2
The inflation rate can be sustained as an equilibrium if
G( ) < C( )
73

or
 1 (k + a )2

1  2 2 2
1 2
2
k + ( )2
<

(
)
2
2 1 + a2
2
Simplifying, this condition becomes





1 + (1 + a2 ) ( )2 2ak + 1 1 + a2 a2 2 k2 < 0

(111)

The questions asks for the minimum value of that can be sustained as
an equilibrium, given the trigger strategy followed by the public in forming their
expectations. So we want to find the minimum value of that satisfies equation
(111).
Solving the quadratic equation (111) for ,



2ak (2ak)2 4 1 2 (1 + a2 )2 a2 2 k2
=
2 (1 + (1 + a2 ))
Simplifying,



ak 1 1 + a2
=
1 + (1 + a2 )

so the minumum is given by





ak 1 1 + a2
=
1 + (1 + a2 )


Note that if 1 1 + a2 < 0, min < 0 and = 0 is a feasible equilibrium. This is the case shown in Figure 2 which plots G( ) C( ) as a
function of for a = 1, = .5, = .9 and k = .1.
min

9. Assume that nominal wages are set at the start of each period but that
wages are partially indexed against inflation. If wc is the contract base
nominal wage, the actual nominal wage is w = wc + (pt pt1 ) where
is the indexation parameter. Show how indexation affects the equilibrium
rate of inflation under pure discretion. What is the effect on average
inflation of an increase in ? Explain why.
Referring to question 1, consider the same production set up so that employment is again given (in log terms) by the marginal product condition:
l=

1
(w p ln A)
1
74

G-C

Figure 3: Gain minus Cost of Deviating

-0.04

-0.02

0.02

0.04

Figure 4:

75

0.06

0.08

0.1

0.12

Substituting in the new expression for the nominal wage,


l=

1
(wc (1 )p p1 ln A)
1

If we assume the base contract wage wc is set equal to the expected market
clearing level, wc = Ep + ln A (1 )l where l is the fixed supply of labor.
Actual employment is
1
(Ep + ln A (1 )l (1 )p p1 ln A)
1

1
(p Ep)
(Ep p1 )
= l +
1
1

l =

while output is
y

(1 )
(p Ep)
(Ep p1 ) + e
1
1
= y + a(1 ) (p Ep) a (Ep p1 ) + e
= l +

where a = /(1 ). Under discretion, Ep and p1 are taken as givens when


the central bank decides on its current price level. Thus, the important difference
introduced by indexation is that the effect of a price surprise is decreasing in the
indexation parameter .
If we now derive the optimal discretionary policy outcome, using for example
the loss function (8.2) and repeating the analysis that led to equation (8.7), the
parameter a of the text is replaced everywhere with a(1 ) and the inflation
bias will be equal to
a(1 )k ak
Notice that indexation ( 0 < < 1) reduces the average inflation bias. By
reducing the impact of a price surprise on real output, indexation reduces the
incentive to induce an output expansion.
10. Suppose the central banks loss function is given by
V cb =


1 
y yn k)2 + (1 + ) 2
2

If y = yn + a( e ) + e and = m + v, verify that the inflation rate


under discretion is given by equation (8.18).
This problem actually is the same as number 7.

76

11. Beetsma and Jensen (1998): Suppose the social loss function is equal to

1  
V s = E y yn k)2 + 2
2
and the central banks loss function is given by

2 

1 
+ t
V cb = E ( ) y yn k)2 + (1 + ) T
2
where is a mean zero stochastic shock to the central banks preferences,
T is an inflation target assigned by the government, and t is a linear
inflation contract with t a parameter chosen by the government. Assume
that the private sector forms expectations before observing . Let y =
yn + ( e ) + e and = m + v. Finally, assume and the supply shock
e are uncorrelated.
(a) Suppose the government only assigns an inflation target (so t = 0).
What is the optimal value for T ?
(b) Now suppose the government only assigns a linear inflation contract
(so T = 0). What is the optimal value for t?
(c) Is the expected social loss lower under the inflation target arrangement or the inflation contract arrangement?
(Notes: This statement of the problem corrects two typos in the text. Also,
in the text, Beetsma and Jensen is listed as forthcoming. It has now appeared
in the Journal of Money, Credit, and Banking, 30 (3), part 1, August 1998,
384-403.)
(a) It will simplify to treat inflation as the central banks choice variable.
From the link between money growth and inflation ( = m + v), one can
easily obtain the rate of money growth needed to achieve the desired expected
inflation rate.
With only an inflation target, the central banks loss function is
2 

1 
V cb = E ( )( e + e k)2 + (1 + ) T
2
where the relationship between output and surprise inflation has been used to
eliminate y yn . The first order condition for the optimal inflation setting,
taking private expectations as given, is


(112)
( )( e + e k) + (1 + ) T = 0
Taking expectations conditional on the publics information set (which does not
include e or ), k + e T = 0 or
e = T + k

77

Substituting this back into the central banks first order condition (112),
= T + ( )k

1+
e
1+

(113)

To find the optimal value for the inflation target, use (113) to evaluate social
1+
loss, noting that e = k 1+
e:


2 

1
1
+

e + T + ( )k
e
V s = E (1 + )k +
2
1+
1+
Minimizing this respect to the target inflation rate yields the first order condition


1+
e =0
E T + ( )k
1+
or
T = k
This is Svenssons (1997) result: to offset the inflation bias, the inflation target
must be set below the socially optimal inflation rate (equal to zero in this case). If
the inflation term in the social loss function had allowed for a non-zero optimal
inflation rate, by having ( )2 rather than simply 2 in V s , then the optimal
target would have been k.
(b) With a linear inflation contract but no inflation target, the central banks
loss function is

1 
V cb = E ( )( e + e k)2 + (1 + )2 + t
2
and first order condition will be
( )( e + e k) + (1 + ) + t = 0
Solving for inflation,
=

( )(e e + k) t
1+

The public will expect an inflation rate of k t. Hence, the central bank will
deliver an inflation rate of
= ( )k

1+
(1 + )
t
e
1+
1+

(114)

Notice that the response to the supply shock is the same under either the target
(see equation 113) or the contract (see 114).

78

To find the optimal value of t from the governments perspective,


 evaluate
1
1+
e
e:
V , first noting that equation (114) implies = k 1+ t 1+
s

Vs




2
1
+
1
E k
t +
ek
2
1+
1+

2
(1 + )
1+
1
t
e
+ E ( )k
2
1+
1+

The first order condition for the optimal t is





 
+

1
t +
ek
0 = E
k
1+
1+
1+



(1 + )
1+
(1 + )
E
t
e
( )k
1+
1+
1+
Evaluating the expectations, and making use of the assumption that E(e) = 0,
 2 

 2 

1
1

t k k + t
t =0
k
1+
1+
1+
1+
Combining terms, this can be written as

 2 




1+
t + 2 k = 0
1+
or



(1 + ) + 2
k
t=
1 + + 2

(115)

(c) To evaluate the social loss function under the alternative policies, it is
useful to start with the expressions for y yn k and and their variances for
each type of policy. Let subscripts T and c denote the targeting regime and the
contract regime. Then, for the targeting regime,
T = k

2 + 2 2
e E2 = 2 k2 +
e
1+
(1 + )2

and
yT yn k = (1 + )k +

1+
e
1+

which implies
E (yT yn k)2 = (1 + 2 )k2 +

79

1 + 2

(1 + )2

2e

For the contract regime, substitute (115) into (114) to obtain



 

(1 + ) (1 + ) + 2
1+
e
c = ( )k
k
2
1+
1 + +
1+

+ 2
1+
2 (1 + 2 ) 2 2 + 2 2
2
e

E
=
k

=
2k +
2 e
1 + + 2
1+
(1 + )
(1 + + 2 )


yc yn k = 1 +
1 + + 2


k+

1+
e
1+

which implies


E (yc yn k) = 1 +

(1 + +


2
2 )

k2 +

1 + 2

(1 + )2

2e

We can now evaluate social loss under the two policy regimes. For inflation
targeting,




1
1 + 2 2
1 2 2 2 + 2 2
s
2 2
(1 + )k +
e +
e
VT =
k +
2
2
(1 + )2
(1 + )2


1
1 + 2
(1 + 2 ) + 2 k2 +
=
2
2
2 (1 + ) e
while for the inflation contract regime,



2
2

1
1
+

1+
Vcs =
2e
2 k +
2
(1 + )2
(1 + + 2 )


1
2 (1 + 2 ) 2 2 + 2 2
+
k +
e
2 (1 + + 2 )2
(1 + )2

1
1 + 2
2
2
=
k +
+
2
2
1 + + 2
2 (1 + ) e
Subtracting Vcs from VTs yields





2
1 2 
s
s
2
+ (1 + ) +
VT Vc =
k
2
1 + + 2



1 2 2
1
k (1 + )
=
2
1 + + 2
which is always positive if 2 > 0, i.e., if there is any uncertainty about the
central banks preferences. Thus, in the face of preference uncertainty, the linear
inflation contract performs better than a simple inflation target.

80

Chapter 9: Monetary-Policy Operating Procedures


1. Suppose equations (9.1) and (9.2) are modified as follows:
yt = it + ut
mt = cit + yt + vt
where ut = u ut1 + t , vt = v vt1 + t and and are white noise processes (assume all shocks can be observed with a one period lag). Assume
the central banks loss function is E[y]2 .
(a) Under a money supply operating procedure, derive the value of mt
that minimizes E[y]2 .
(b) Under an interest rate operating procedure, derive the value of it
that minimizes E[y]2 .
(c) Explain why your answers in (a) and (b) depend on u and v .
(d) Does the choice between a money supply procedure and an interest
rate procedure depend on the i s? Explain.
(e) Suppose the central bank sets its instrument for two periods (for
example, mt = mt+1 = m ) to minimize E[yt ]2 + E[yt+1 ]2 where
0 < < 1. How is the instrument choice problem affected by the
i s?

a) Under a money supply procedure, the money demand relationship implies


that the interest rate is it = c1 (yt + vt mt ). Output is then equal to
yt

(yt + vt mt ) + ut
c
(mt vt ) + cut
c+

=
=

(116)

The objective is to pick mt to minimize E[y]2 . The first order condition is


(mt vt ) + cut
2
E
=0
c+
c+
Since the shocks are assumed to be observed with a one period lag, E (vt ) =
v vt1 and E (ut ) = u ut1, so this first order condition requires that mt satisfy
mt v vt1 + cu ut1 = 0
or
mt = v vt1
81

c

u ut1

The money supply is adjusted to offset the forecasted effects of vt and ut on


output:
yt =

ct t
c+

(117)

b) Under an interest rate procedure, output is equal to


yt = it + ut

(118)

and it is chosen to minimize E[y]2 = E[it + ut ]2 . The first order condition


is
2E[it + ut ] = 0
or
it =

u ut1

The interest rate is adjusted to offset the predicted aggregate demand shock, while
money demand shocks do not affect output and so do not require any adjustment
in the interest rate instrument.
c) As noted under parts (a) and (b), the optimal policy will involve trying to
insulate output from the two shocks ut and vt . If these could
 c  be observed before
=
v

policy is set, the optimal


policies
would
be
m
t
t
ut under a money
 
procedure and it = 1 ut under an interest rate procedure. Under certainty
equivalence (which holds in the linear model with a quadratic objective function),
the optimal policy simply replaces ut and vt with the best forecast of the shocks,
u ut1 and v vt1 .
d) The loss function under the m procedure is
 
2

 
v vt1 c u ut1 vt + cut
L(m) = E
c+

2
c(ut u ut1 ) (vt v vt1 )
= E
c+

2
ct t
= E
c+
which is independent of both u and v .
Under the interest rate procedure, the loss is
L(i) = E [ut u ut1 ]2 = E [t ]2
which is also independent of u and v . Consequently, the comparison between
a money procedure and an interest rate procedure will not depend on either u
82

or v . Since the predictable component of the shocks (the component that does
depend on u and v ) is offset under both policies, the comparison will only
depend on how well the different policies insulate output from the unforecastable
shocks ( t and t ).
e) If the objective is to set m or i at time t for two period to minimize
E[yt ]2 + E[yt+1 ]2 , the analysis becomes more complicated and the comparisons
between the money supply and the interest rate policies will depend on the serial
correlation properties of the shocks. Starting with the money supply procedure,
we can use (116) for output to write the loss function as

2
(m vt+1 ) + cut+1
(m vt ) + cut
+ E
E
c+
c+
since, by assumption, m is fixed for two periods. The first order condition is



2
(m vt ) + cut
(m vt+1 ) + cut+1
E
+ E
=0
c+
c+
c+
From the process followed by the disturbances, E(ut ) = u ut1 and E(ut+1 ) =
2u ut1 , while E(vt ) = v vt1 and E(vt+1 ) = 2v vt1 . Using these in the first
order condition, the optimal m must satisfy


 
(m v vt1 ) + cu ut1 + m 2v vt1 + c2u ut1 = 0
or
m =

(1 + v )v vt1 c(1 + u )u ut1


(1 + )

Since m is fixed for two periods, it adjusts to offset what amounts to the average
discounted expected shocks over the two periods. As a consequence, output will
not be perfectly insulated from the forecasted components of ut , ut+1 , vt , or
vt+1 :








u
v
u ut1 vt 1+
v vt1
c ut 1+
1+
1+
yt =
c+
c [(1 + ) t + (1 u ) u ut1 ] [(1 + ) t + (1 v ) v vt1 ]
=
(1 + ) (c + )



ct t

c(1 u )u ut1 (1 v )v vt1


=
+
(119)
c+
1+
c+
(which should be compared with equation 117) and








1+v
u

v
c ut+1 1+
t1
t+1
t1
u
v
1+
1+
yt+1 =
c+




c (1 + ) t+1 + u t (1 u ) u ut1
=
(1 + ) (c + )




(1 + ) t+1 + v t (1 v ) v vt1

(1 + ) (c + )
83

or
yt+1



 


c t+1 + u t t+1 + v t
c (1 u ) u ut1 (1 v ) v vt1

=
c+
(1 + ) (c + )

Forecast errors made in period t, and therefore not fully offset, continue to affect
output in period t + 1 if the disturbances are serially correlated ( u t and v t
show up in the expressions for yt+1 ).
Under an interest rate policy, the objective is to pick i to minimize
2

E [i + ut ] + E [i + ut+1 ]
so the first order condition is

2 {E [i + ut ] + E [i + ut+1 ]} = 0
i + u ut1 i + 2u ut+1 = 0
or
i =

(1 + u )u ut1
(1 + )

and output in the two periods will equal




1 + u
(1 u )
u ut1
yt = ut
u ut1 = t +
1+
1+
and
yt+1

(1 + u )u ut1
(1 + )

 2
(1 + ) u ut1 + u t + t+1 (1 + u )u ut1
=
(1 + )
(1 u )
ut1
= t+1 + u t
(1 + ) u
= ut+1

Since the variance of output under the two policies will now depend on u
and u , the comparison of the loss functions under the two policies will no
longer be independent of the serial correlation properties of ut and vt . For
example, suppose u = 0 but v = 0. Under an interest rate policy, output does
not depend on the v disturbances, so yt = t and yt+1 = t+1 , but under the
money supply rule, equation (119) becomes




ct t

yt =
(1 v )v vt1
c+
1+
c+
84

and


yt+1 =

ct+1 t+1
c+


(1 v ) v vt1

v t +
c+
(1 + ) (c + )

which still depends on v . Since 2v = 2 /(1 2v ), the variance of output under


a money supply procedure is



2
2 
c

(1 v )2 2v
2
2
2
+
1 + v +
2
E[yt ]m =
c+
c+
(1 + )2 (1 2v )



2
2 
2
c

(1

v
v
=
2 +
1 + 2v +
2
c+
c+
(1 + )2 (1 + v )
while under an interest rate procedure,
E[yt ]2i = 2

2. Solve for the i s appearing in (9.11) and show that the optimal rule for
the base is the same as that implies by the value of given in (9.10).
The i s that appear in equation (9.11) are obtained by calculating the least
squares forecast of each shock based on the observed value of the interest rate.
For the model of section 9.3.2, the equilibrium expression for the interest rate
is given by equation (9.9):
i=

v+u
+c++h

We can use this to calculate the forecasts of v, , and u, conditional on observing i. In the text, these forecasts are denoted v,
, and u
(see page 394).
From the
least
squares
formula,
the
forecast
of
a
variable
y,
conditional
on x,



2
x
where

is
the
covariance
between
x
and
y
and

is
the
is y = x,y
2
x,y
x
x
variance of x. (This assumes both y and x have zero means.)
Applying this formula, we have


v,i
v =
i
2

i 2


v
2v
+c++h
i
= 2 +2 +2 i = ( + c + + h)
v

u
2v + 2 + 2u
2
(+c++h)

so


v =

+c++h
2v + 2 + 2u
85

2v

Similarly,

=
and


u =

+c++h
2v + 2 + 2u

+c++h
2v + 2 + 2u

2u

The next step is to substitute these expressions into the policy rule (9.11):


c+h
u + v i
b =




+c++h
c+h 2
2
2
+ v + i
=

2v + 2 + 2u
u
or, since b = i,

+c++h
2v + 2 + 2u



c+h 2
2
2
+ v +

Solving this for yields




2v + 2
= (c + h) +
2u

which proves that the policy rule (9.11) yields the same response to the interest
rate as was found in equation (9.9), and the optimal policy rule is

 2


v + 2
b = (c + h) +
i = i
2u
3. Suppose the money demand relationship is given by m = c1 i + c2 y +v.
Show how the choice of an interest rate versus a money supply operating
procedure depends on c2 . Explain why the choice depends on c2 .
Using the basic model given by equation (9.1) with the money demand equation specified in the problem, interest rates and output under a money supply
procedure are given by
 
1
i=
(c2 y m + v)
c1
and

=
(c2 y m + v) + u
c1


1
=
[a(m v) + c1 u]
c1 + c2
86

and the loss function E (y)2 is minimized when








(m v) + c1 u

m
2
E
=2
E
=0
c1 + c2
c1 + c2
c1 + c2
c1 + c2
or m = 0. The loss function is then equal to

L(m) =

1
c1 + c2

2

 2 2

v + c21 2u

In contrast, under an interest rate procedure, y = i + u, the variance of


output is minimized if i = 0, and the loss function then takes on the value
L(i) = 2u
An interest rate rule is preferred if L(i) < L(m), or if


2c1
2v > c2 +
c2 2u

which should be compared with equation (9.6) on page 390 of the text.
A large value of c2 (a large income elasticity of money demand) makes
it more likely that a money supply procedure will be preferred. Consider the
impact on output of a positive u shock. If c2 is large, the resulting rise in
output has a large impact on money demand. This in turn causes interest rates
to rise, offsetting the original rise in output. Thus, u shocks have a smaller
impact on output under an m procedure when c2 is large. Similarly, a positive
v that increases money demand and raises interest rates under an m procedure,
will lower output but the decline in y has, when c2 is large, a strong impact in
lowering money demand. As a result, interest rates need to rise less to maintain
money market equilibrium after the positive v shock, and, with a smaller rise in
i, y falls less.
These results can be illustrated by Figure 1. The negatively sloped solid line
is the IS equation y = i when u takes on its expected value of 0. The
positively sloped lines AA and BB give money market equilibrium when m= 0
and v takes on its expected value of 0 (that is, these lines show i = cc21 y).
The line BB is drawn for a larger value of c2 . The dotted negatively sloped line
shows the results of a positive u shock; output rises less when c2 is large; output
rises from he the level associated with point C to D if c2 is small, but it rises
only to E when c2 islarger. A positive v shock shifts AA and BB to A A and
B  B  (since i = c11 (c2 y + v), the vertical shift is the same for both). Again,
output is less affected when c2 is large, falling only from C to F when c2 is
large, rather than C to G.

87

Figure 5: Chapter 9, Problem 3 The impact of c2 under a money supply


operating procedure

B'
B

Interest Rate

A'

A
G
E
F
D
A'

A
B'

Output

4. Prices and aggregate supply shocks can be added to Pooles analysis by


using the following model:
yt = yn + a( t Et1 t ) + et

(120)

yt = yn (it Et t+1 ) + ut

(121)

mt pt = 0 it + yt + vt

(122)



Assume the central banks objective is to minimize E y 2 + 2 , and that
are disturbances are mean zero, white noise processes. Both Et1 t and
the policy instrument must be set prior to observing the current values of
the disturbances.
(a) Calculate the expected loss function if it is used as the policy instrument. (Hint: Give the objective function, the instrument will always
be set to ensure expected inflation is equal to zero.)
(b) Calculate the expected loss function if mt is used as the policy instrument.
(c) How does the instrument choice comparison depend on
88

i. the relative variances of the aggregate supply, demand, and money


demand disturbances?
ii. the weight on stabilizing output fluctuations ?
Note: There is a typo in this problem; the loss function should be


2
E (y yn ) + 2
(or one can simply assume yn = 0 as a normalization).
a) Under an interest rate policy, the money demand equation given by (122)
is not needed. Using the hint and setting Et1 t = Et t+1 = 0, equation
(140) implies yt yn = it + ut . Using this in (120), inflation will equal
t = a1 (it + ut et ). This means we can write the policy problem in terms
of the policy instrument it as


2 
1
2
(it + ut et )
min E (it + ut ) +
it
a
The first order condition is


1
2E (it + ut ) + (it + ut et ) = 0
a

(123)

The problem specified that the policy instrument must be set before observing
the disturbances, so in evaluating the first order condition, E (ut ) = E(ut ) =
E (et ) = 0. Thus, (100) becomes it a1 it = 0 or
it = 0
With this setting for the nominal interest rate,
yt yn = it + ut = ut

(124)

and
t =

1
u t et
(it + ut et ) =
a
a

Notice that under a policy that sets i = 0, inflation is a mean zero, serially
uncorrelated process, so Et1 t = Et t+1 = 0 as was assumed.
Using these results, the expected loss under an interest rate policy,
L(i) =

(1 + a2 ) 2u + 2e
a2

(125)

b) Under a money rule, we need to use equation (122), solving it for the
nominal interest rate and then use this result to eliminate it from equations
(120) and (121). Since equations (120) and (140) are expressed in terms of
89

the rate of inflation while (122) involves the price level, we can either express
inflation as pt pt1 , and then solve for the price level, or, since pt1 is known
when policy is set, we could replace mt pt in (122) with mt + pt1 t and
solve for the rate of inflation. Since the loss function is expressed in terms of
inflation, this latter approach is more convenient.
From (122), the nominal rate of interest is
it =

c0 + t mt + pt1 + yt + vt
c

(126)

Substituting this into (140), and using the earlier hint about expected inflation,
we have


c0 + t mt + pt1 + yt + vt
yt yn =
+ ut
c
(mt c0 t pt1 yn vt ) + cut
=
c+
which can be solved jointly with (120) to yield
a (mt c0 pt1 yn vt ) + acut + et
a (c + ) +

(127)

(mt c0 pt1 yn vt ) + cut (c + ) et


a (c + ) +

(128)

yt yn =

t =

Now substitute these two solutions into the loss function. The policy problem is
then


2 
a (t vt ) + acut + et
(t vt ) + cut (c + ) et
min E
+
t
a (c + ) +
a (c + ) +
where, for convenience, t has been defined as mt c0 pt1 yn and can be
viewed as the policy instrument. The first order condition for the choice of t
is




2
at
t
a
+
=0
a (c + ) +
a (c + ) +
a (c + ) +
where we have used the fact that at the time policy is chosen, Eut = Eet =
Evt = 0. The first order condition is satisfied for t = 0, or
mt = c0 + pt1 + yn
Using (127) and (128), output and inflation under a money instrument are
yt yn =

acut avt + et
a (c + ) +
90

and
t =

cut vt (c + ) et
a (c + ) +

and the loss function is



 


1 + a2 c2 2u + 2 2v + 2 + (c + )2 2e
L(m) =
2
[a (c + ) + ]

(129)

c) The instrument choice hinges on a comparison of the loss L(i) given


in (125) and the loss L(m) given in (129), with an interest rate instrument
preferred if
L(i) < L(m)
or if
2

(1 + a

)2u
a2

2e

<


 


2
1 + a2 c2 2u + 2 2v + 2 + (c + ) 2e
[a (c + ) + ]2

(130)

This can be rewritten with some rearranging as implying L(i) < L(m) if


2


(a (c + ) + ) a2 c2
2a (c + ) + (1 a2 ) 2
2
2
e
u +
v >
a2 2
(1 + a2 )
This shows that the comparison depends on the different variance terms. A
money oriented operating procedure is less likely to be desirable if money demand
shocks are large (i.e., 2v is large). The coefficient on 2u is positive, so an
interest rate rule is more likely to be preferred if aggregate demand shocks are
important (i.e., 2u is large), while it is also likely to be preferred if supply
disturbances are large (i.e., 2e is large). Notice that if 2e is zero (no supply
shocks), the comparison is independent of the preference weight .
The weight on stabilizing output fluctuations, , affects the comparison only
if 2e > 0. In the absence of aggregate supply shocks, there is no conflict in
this model between stabilizing output and stabilizing inflation, so the operating
procedure comparison will be independent of . When aggregate supply shocks
are present ( 2e > 0), then there can be conflicts between stabilizing output and
stabilizing inflation. If output objectives are very important ( large), then it
is more likely an interest rate procedure will be preferred. To understand why,
consider what happens in the face of a positive aggregate supply shock. Under an
interest rate procedure, aggregate demand remains constant (see equation 124),
so output is stabilized and inflation must fall. Such a policy will be preferred
if is large. Under a money supply procedure, in comparison, output will rise
and inflation will fall. Since both adjust, inflation falls by less than under the i
policy. A policy maker who cares more about inflation stabilization (i.e., has a
lower ) will prefer money supply operating procedure.

91

5. Using the intermediate target model of section (9.3.3) and the loss function
(9.15), rank the policies that set it equal to t , iTt , and t + xt .
The basic model of section 9.3.3 consists of the following equations:
yt = a(t Et1 t ) + zt
yt = (it Et t+1 ) + ut
mt pt1 t = yt cit + vt
These appeared as equations (9.12) - (9.14) on page 397. The loss function
(9.15) is
2

V = E ( )
The first policy to evaluate sets

 
1
it = t = +
(u ut1 z zt1 )

(see equation 9.17, page 397). The value of the loss function under this policy
was given at the bottom of page 397:
 2

1  2
+ 2e
(131)
V (t ) =
a
Under the second policy,
it = iTt = t +

(1 + a)t et + at
ac + (1 + a)

(see equation 9.21 on page 399). The inflation rate is (see page 399)
 
c ( + c)et t
t iTt = + t
ac + (1 + a)
and the value of the loss function under this policy was given in the middle of
page 400:

2
 2 2

1
T
c + ( + c)2 2e + 2 2
(132)
V (it ) =
ac + (1 + a)
Comparing (131) and (132),

 2
2

 2 2

1
1  2
T
2
+ e
c + ( + c)2 2e + 2 2
V (t ) V (it ) =
a
ac + (1 + a)



2ac(1 + a) + 2 (1 + a)2 2 + 2a2 (c + ) 2e
=
a2 (ac + (1 + a))2

2
1
2 2

ac + (1 + a)
92

The first term is positive, indicating that the intermediate targeting rule leads to
a smaller loss ( V (t ) > V (iTt )) if the only disturbances are demand and supply
shocks ( and e). As discussed in the text, however, the intermediate targeting
procedure can do worse if money demand shocks are important ( V (t ) < V (iTt )
if 2 is large).
The final policy sets it equal to t + xt , where xt is defined below equation
(9.22) on page 401 as

 

1
1
xt = 1 +
t
et + t
a
a
and was defined, also on page 401, as

 
a(1 + a) 2 + a2e
1

(1 + a)2 2 + 2e + a2 2
Using the definition of t , the rate of interest under this policy is
 
 



1
1
1

it = t + xt = +
1+
(u ut1 z zt1 ) +
t
et + t

a
a
To evaluate the loss function under this policy, use equation (9.16) of the
text to find the equilibrium rate of inflation when it = t + xt :
(a + ) (t + xt ) + ut zt
a
(ut u ut1 ) (zt z zt1 ) xt

= +
a
 
1

= +
(t et xt )
a

(t + xt ) =

Problem 2 showed that the value of was related to the best forecasts of and
e, conditional on observing x. We can write inflation under this policy as
 
1
(t + xt ) = +
{[t E(t | x)] [et E(et | x)]}
a
 
Comparing this to the policy that lead to V (t ), in which = a1 (t et ),
it is clear that the variance of inflation around will be smaller with the
t + xt policy since the variance of [t E(t | x)] is less than or equal
to the variance of , and similarly for the comparison of the variances of
[et E(et | x)] and et . Therefore,
V V (t + xt ) V (t )
To compare the loss under the intermediate target policy iT and the policy
that optimal uses information (t + xt ), use the equation near the bottom of
page 400 to write
iTt = t + T xt
93

Using equation (9.16). inflstion can be written as




(a + ) t + T xt + ut zt
T
t (it ) =
a
t et T xt

= +
a

 
T
so the loss function V (i ) is equal to the variance of a1 t et T xt .
Inflation around under the t + xt policy is
t et T xt
a
1

and the loss function V is the variance of a (t et xt ). But since


was chosen to minimize this variance, it must be that V V (iT ).
t (t + xt ) = +

6. Show that if the nominal interest rate is set according to (9.17), the expected value of the nominal money supply is equal to m
given in (9.19).
According to equation (9.17) on page 397, the interest rate takes the value
 
1

it = +
(u ut1 z zt1 )

With inflation given by equation (9.18) and output by (9;12), the money demand
equation (9.14) can be written as
mt pt1

= t + a (t ) + zt cit + vt


c
t et
(u ut1 z zt1 ) + vt
= + (1 + a)
+ zt c
a

since E t1 t = . Taking expectations as of time t 1 of this equation and


solving for E t1 mt ,
c

c
u ut1 + 1 +
zt1 + v vt1
Et1 mt = (1 c) + pt1

z
which is the same as the expression for m
in equation (9.19).
7. Suppose the central bank is concerned with minimizing the expected value
of a loss function of the form
L = E[T R]2 + E[if ]2
which depends on the variances of innovations to total reserves and the
funds rate ( is a positive parameter). Using the reserve market model of
section 9.4.2, find the values of d and b that minimize this loss function.
Are there conditions under which a pure nonborrowed reserves or a pure
borrowed reserves operating procedures would be optimal?

94

The reserves market model from section 9.4.2 consists of a total reserves
demand equation, a borrowed reserves demand equation, a supply of nonborrowed
reserves equation, and an equilibrium condition. These are specified as
T R = if + vd
BR = b(if id ) + v b
NBR = d vd + b v b + v s
and the equilibrium condition that
T R = BR + N BR
= if + b(if id ) + (d 1)vd + (1 + b )v b + v s
Substituting these first three equations into the equilibrium condition and
solving for the funds rate if yield



1
bid (1 + b )vb + (1 d )vd vs
if =
a+b
If the objective is to pick d and b to minimize L = E[T R]2 + E[if ]2 , the
first order conditions will be
 f
 f
i
i
f
]
2E[if + vd ]
+
2E[i
=0
(133)
d
d
and


2E[if + vd ]

if
b


+ 2E[if ]

if
b


=0

(134)

 f
 f
i
i
= v d /(a + b), while
= vb /(a + b).
To evaluate these, note that
d
b
Hence, equation (133) implies




vd
vd
0 = E[if v d ]
+ 2E[if ]
a+b
a+b






d

1 d
1
2
2
2
d +
2d
=
d
(a + b)2
a+b
(a + b)2
Solving for d ,
(1 d ) =

95

(a + b)
2 +

d = 1

(a + b)
2 +

(135)

Equation (134) yields








(1 + b )vb
v b
(1 + b )v b
vb
E
+ E
=0
a+b
a+b
a+b
a+b
or


2

1 + b
(a + b)2


2b

1 + b
(a + b)2


2b = 0

which can be solved for the optimal b , yielding


b = 1

(136)

To understand these results, start first with the b = 1 finding. A shock


to borrowed reserve demand should generate an equal but opposite movement
in nonborrowed reserves. This keeps total reserves unchanged. Since neither
total reserve supply or demand have changed, the funds rate is left unchanged.
Thus, setting b = 1 and accommodating shifts in borrowed reserve demand
completely insulates both T R and if from vb shocks.
From (135), the optimal value of d is equal to 1 (a+b)
2 + is less than 1 and
depends on the preference parameter . In response to a shock to total reserve
demand, reserve supply adjusts to fully accommodate the shift if d = 1; this
would succeed in insulating the funds rate from the shock, but it would lead total
reserves to move one-for-one with vd . By setting d < 1, reserve supply less
than fully accommodates the shift in reserve demand. This means the funds rate
rises (falls) if vd > 0 ( < 0). This means the funds rate moves more, but total
reserves move less, and this will be optimal since the policy maker cares about
both E[T R]2 and E[if ]2

96

Chapter 10: Interest Rates and Monetary Policy


1. Suppose (10.1) is replaced by a Taylor sticky price adjustment model of the
type studied in Chapter 5. Is the price level still indeterminate under the
policy rule (10.5)? What if prices adjust according to the Fuhrer-Moore
sticky inflation model?

Equation (10.1) relates the level of real output to the price surprise term
pt Et1 pt ; the actual level of prices doesnt really matter. With a Taylor-type
price adjustment model of the type discussed in section 5.5.1, the price level at
time t will depend on prices in previous periods. Employing a simple version
in which prices are a markup over wages, and nominal wages are set for two
periods with half of all wages set each period, the aggregate price level in period
t will be equal to
pt =

1
(xt + xt1 )
2

(137)

where xt is the contract wage set in period t. If wage setting depends on the
expected price level over the two periods the wage is set and on the current state
of economic activity,
xt =

1
(pt + Et pt+1 ) + kyt
2

(138)

as in equation (5.44) on page 216. Substituting (138) into (137),


pt =

1
1
(pt + Et pt+1 + pt1 + Et1 pt ) + k (yt + yt1 )
4
2

Multiplying both sides by 4 and rearranging, the Taylor adjustment model implies
pt =

1
2
(pt1 + Et pt+1 + Et1 pt ) + k (yt + yt1 )
3
3

If we combine this with the IS equation (10.7) and Fisher equation (10.8), then
under an interest rate peg, equilibrium is obtained as the solution to
pt =

1
2
(pt1 + Et pt+1 + Et1 pt ) + k (yt + yt1 )
3
3

(139)

yt = 0 1 rt + ut

(140)

iT = rt + (Et pt+1 pt )

(141)

To see if the price level is determinate, consider what would happen if, at the end
of period t1, the public expected the price level in all future periods to be higher
97

by %. Since the model is specified in log form, we need to check whether adding
to pt , Et1 pt and Et pt+1 would affect the equilibrium. Clearly, equations
(140) and (141) would be unaffected, (140) because it does not involve the price
level, and (141) because expected inflation is also unaffected if the price level
jumps by % and remains at this new higher level: (Et pt+1 + (pt + )) =
Et pt+1 pt . But equation (139) is affected; pt1 is predetermined it cant
jump when expectations change. So when pt , Et pt+1 and Et1 pt increase by
, the left side of 139) goes up be , while the right side only goes up by 23
because pt1 cant increase by ; is no longer satisfied if pt jumps to pt + .
Equilibrium is determined by the historical price level.
If a Fuhrer-Moore model of inflation adjustment is used, then equations (140)
and (141) remain unchanged, but the inflation process is different. From equation (5.61), the change in the contract wage is given by
xt =

1
( t + Et t+1 ) + 2kyt
2

and inflation is equal to


t

=
=

1
(xt + xt1 )
2
1
(t + Et t+1 + t1 + Et1 t ) + k (yt + yt1 )
4

Rearranging,
t =

1
4
[ t1 + Et t+1 + Et1 t ] + k (yt + yt1 )
3
3

In terms of pt , this can be written as


pt = pt1 +

1
4
[t1 + Et t+1 + Et1 t ] + k (yt + yt1 )
3
3

(142)

Again. history pins down the price level. A jump in pt and all future expected
price levels leaves Et t+1 and Et1 t unaffected. Lagged inflation t1 is also
unaffected since it is predetermined as of time t. But pt also depends on the level
of past prices pt1 so a % jump would not leave the equilibrium unaffected.
Both the Taylor and the Fuhrer-Moore models imply that the current price
level depends, in part, on the previous price level. Thus, the price level is determinate if we can take the historical value of pt1 as given.
2. Derive the values of the unknown coefficients in (10.10) and (10.11) if the
money supply process is given by (10.15).
Equations (10.10) and (10.11) were used to derive the equilibrium processes
followed by the price level and the nominal interest rate when the nominal money

98

supply was set according to (10.9). Suppose instead that mt is determined by


(10.15), repeated here as


mt =  + 0 t + it iT
The rest of model is given by equations (10.1) - (10.4). Using this new process
for mt , the proposed solutions for pt and it will need to be modified to allow
for the possibility that either the price level or the nominal rate of interest, or
both, may be affected by the deterministic trend that appears in the money supply
process. Thus, we need to consider the solutions
pt = b10 + b11 mt1 + b12 et + b13 ut + b14 vt + b15 t

(143)

it = b20 + b21 mt1 + b22 et + b23 ut + b24 vt + b25 t

(144)

Combining (10.1), (10.2), and (10.4) yields (10.12) for the nominal rate of
interest:
it =

0 yc
1

[a (pt Et1 pt ) + ut et ] + Et pt+1 pt


1
1

(145)

while (10.1), (10.3), and the new policy rule (10.15) yield


pt =  + 0 t + it iT + cit y c a (pt Et1 pt ) et vt
=  iT yc + 0 t + ( + c)it a (pt Et1 pt ) et vt (146)
which can be compared to (10.13), obtained using the policy (10.9) in which the
money supply depended on lagged money.
We need to solve for the unknown coefficients in (143) and (144) such that
equations (145) and (146) are satisfied for all realizations of the e, u, and v
and all t.
Using the proposed solutions,
pt Et1 pt = b12 et + b13 ut + b14 vt
and
Et pt+1

= b10 + b11 mt + b15 (t + 1)


= b10 + b11  + b15 iT + (b11 0 + b15 ) t + it + b15 (t + 1)

Using these expressions for the expectational terms, together with the proposed
solutions, we can determine the unknown coefficients. For example, consider
the coefficients b11 and b21 on mt1 From (145), these must satisfy
b21 = b11
while from (146), they must satisfy
b11 = ( + c)b21
99

or
b11 = b21 = 0
Proceeding in a similar manner, the coefficients b12 and b22 on et must
satisfy
b22 =

a
1
b12 b12
1
1

and
b12 = ( + c)b22 ab12 1
or

b12

1 + + c
=
1 (1 + a) + ( + c)(1 + a)


b22 =

1 1
1 (1 + a) + ( + c)(1 + a)

For the coefficients on ut :


b23 =

a
1
b13 b13 +
1
1

and
b13 = ( + c)b23 ab13
or


b13 =

b23 =

+c
1 (1 + a) + ( + c)(1 + a)
1+a
1 (1 + a) + ( + c)(1 + a)

For the coefficients on vt :


b24 =

a
b14 b14
1

and
b14 = ( + c)b24 ab14 1
or


b14 =

1
1 (1 + a) + ( + c)(1 + a)
100


b24 =

1 + a
1 (1 + a) + ( + c)(1 + a)

For the trend,


b25 = b15 b15 = 0
and
b15 = 0 + ( + c)b25 = 0
Finally, for the constants


b20


0 y c
=
+ b10 + b15 b15
1


0 y c
=
+ b10
1

b10 =  + iT +

( + c) (0 y c )
+ ( + c)0 y c
1

Collecting these results, the equilibrium processes for the nominal interest
rate and the price level are (ignoring the constant terms),
it = b20 +

(1 1) et + (1 + a) ut + (a + 1 )vt
1 (1 + a) + ( + c)(1 + a)

pt = b10 + 0 t +

( + c) ut (1 + + c) et 1 vt
1 (1 + a) + ( + c)(1 + a)

These can be compared to the solution coefficients reported on page 437 and the
interest rate solution given in equation (10.14).
3. Suppose the money supply process in section 10.4.2 is replaced with
mt = mt1 + qt1 + t
so that the policy maker is assumed to response with a lag to the real
rate shock, with the parameter viewed as a policy choice. Thus, policy
involves a choice of and , with the parameter capturing the systematic
response of policy to real interest rate shocks. Show how the effect of qt
on the one and two period nominal interest rates depends on . Explain
why the absolute value of the impact of qt on the spread between the long
and short rates increase with .

101

This problem, and the following one, both use the model of section 10.4.2;
they differ in terms of the process followed by the nominal stock of money. It
will be convenient to solve the model for a more general specification of mt ,
allowing one to then obtain the solutions for Problems 3 and 4 (as well as the
results in the text) as special cases.
The model consists of the following four equations (see equations 10.29 10.31 of the text):
Rt = qt
Rt =

1
[it Et t+1 + Et it+1 Et t+2 ]
2

(147)
(148)

mt pt = ait + vt

(149)

mt = mt1 + qt1 + t t1

(150)

and

where Rt is the two-real period interest rate, qt is an exogenous real rate shock,
it is the one period nominal rate, is the inflation rate, the third equation is a
money demand relationship, and the money supply process used in section 10.4.2
(equation 10.32) has been replaced by the one specified in the question. Notice
that the case considered in the text had = = 0, Problem 3 considers the case
with = 0, and Problem 4 sets = 1, = 0 (and notice that to distinguish
between the coefficient on lagged money and that on the lagged shock ( t1 ) I
have renamed the latter for the purposes of deriving the general solution; in
Problem 4, mt has a coefficient of 1 and the coefficient on t1 is called ).
To answer this question, we need to solve for the one and two-period nominal
rates, together with the interest rate spread,
st I t it =

1
(Et it+1 it )
2

Using (149) to eliminate the one-period nominal rate from equation (148),
the price level prices must satisfy

1 pt mt + vt
(Et pt+1 pt )
qt =
2
a



pt+1 mt+1 + vt+1
+Et
Et (pt+2 pt+1 )
a
or
2aqt = pt mt + vt + apt + Et pt+1 Et mt+1 aEt pt+2
Solving for pt ,
(1 + a)pt = 2aqt + mt vt Et pt+1 + Et mt+1 + aEt pt+2
102

Now use (150) to obtain


Et mt+1 = mt + qt t
so that
(1 + a)pt = 2aqt + (1 + )mt vt + qt t Et pt+1 + aEt pt+2

(151)

from which we can guess that the solution for pt is of the form
pt = b1 mt + b2 qt + b3 vt + b4 t
Using this to evaluate (151),
(1 + a) (b1 mt + b2 qt + b3 vt + b4 t ) = 2aqt + (1 + )mt b1 (mt + qt + t )
+ab1 (mt + qt t ) vt + qt t
since Et pt+1 = b1 Et mt+1 = b1 (mt + qt t ) and Et pt+2 = b1 Et mt+2 =
b1 Et mt+1 = b1 (mt + qt t ). This equilibrium expression holds for all
realizations of mt and the random disturbances if4
b1 =

b2 =

1
1+a

1
1 + a(1 )


2a +

b3 =
and


b4 =

or

pt =

a
1+a

a
1 + a(1 )

1
1+a



1 + a(1 )







1
1
a
a
mt +
2a +
qt vt

1 + a(1 )
1+a
1 + a(1 )
1 + a(1 ) t
(152)

which can be compared to equation (10.34) of the text. The case in the text is
obtained by setting = = 0.
Now that we have the solution for the price level, the one-period nominal
interest rate is, from (149),
it

4 In

1
(pt mt + vt )
a




1
1

=
mt +
2+
qt
1 + a(1 )
1+a
1 + a(1 )





1
1

t +
vt
1+a
1 + a(1 )
1+a

deriving the solution for b1 , the fact that 1 2 = (1 + )(1 ) is used.

103

(153)

This implies that




Et it+1


1
=
Et mt+1
1 + a(1 )


1
=
(mt + qt t )
1 + a(1 )

The two period rate is


It

1
[it + Et it+1 ]
2 


 
1
1
1 2
=

mt +
vt
2
1 + a(1 )
1+a




1
1
( a(1 ))
+
2+
qt
2
1+a
1 + a(1 )



1
1
( a(1 ))

t
2 1+a
1 + a(1 )

(154)

while the spread is


St

=
=

1
(Et it+1 it )
2 






1
1
1
(1 )2
2 + a(1 )
mt
vt +
t
2
1 + a(1 )
1+a
1+a
1 + a(1 )

 


1
(2 + a(1 ))

(155)
2+
qt
1+a
1 + a(1 )

For the specific money process assumed for Problem 3, = 0. Thus, the
terms involving t all become equal to 0. Interest rates depend on because
when differs from zero, agents will adjust their forecast of the future money
supply once they observe qt . Suppose > 0; the money supply is increased
in response to a positive shock to the real rate of interest. Then a positive q
realization causes an upward revision in the future money supply and the future
price level. This raises expected future inflation and so the one-period rate it
rises more than in the = 0 case (see equation 153). With 0 < < 1,
the expected future money supply returns only gradually to its baseline after a
positive q shock. From (155), a positive q shock lowers the spread, but the
To understand why, consider the
absolute value of the impact increases with .
case in which = 1 so that the effect of q on m is permanent. Observing q > 0
raises Et mt+1 and Et t+1 , contributing to the rise in it . But since the money
supply is now expected to remain at this higher level, Et t+2 is unchanged. The
long rate rises less than the short rate and the difference between the two is
larger if qt has a large impact on mt+1 (i.e., if is large). When | |< 1, the
initial rise in mt is expected to gradually be reversed, so Et t+1 will actually
fall, increasing it relative to Et it+1

104

4. Suppose the money supply process in section 10.4.2 is replaced with


mt = mt1 + t t1
Does it depend on ? Does It ? Explain.
We can used equations (153), (154), and (155) from Problem 3 to answer
this question, simply modifying the parameters to reflect the new money supply
process. In particular, the coefficient on lagged money is now equal to 1 (this
coefficient was in Problem 3) and the coefficient on the lagged real rate shock
is now zero ( = 0). Finally, the coefficient on t1 ,which was called in
Problem 3, is renamed in the current Problem.
With these changes, the solutions for the one-period nominal rate and the
two-period nominal rate become


1
(156)
it =
(2qt + vt t )
1+a
1
It =
2

1
1+a


[2qt + vt t ]

(157)

The one-period nominal rate does depend on ; a positive realization of t


increases the period t money supply. If this increase were permanent, the expected rate of inflation would not be affected as the current and expected future
price levels would rise in proportion to the increase in mt . With nonzero, however, some of the change in mt is offset in period t + 1 ( Et mt+1 = mt t ).
If 0 < < 1, for example, the money stock is expected to be lower in t + 1 than
in period t. This reduces expected inflation and the nominal rate falls.
The two-period nominal rate is
It =

1
1
(it + Et it+1 ) = it
2
2

since Et it+1 = 0. The future one-period rate is unaffected (since the money
supply remains constant at its t + 1 value Et mt+2 = mt t ), so the twoperiod rate moves half as much as the one-period rate.
5. Show that equation (10.43) implies rt =

1
1+D

 
i=0

D
1+D

i

Equation (10.43) is repeated here:


rt D [Et rt+1 rt ] = ift Et t+1
which can be written as


  D 
1
ift Et t+1 +
rt =
Et rt+1
1+D
1+D
105



Et ift+i t+1+i .

Updating this one period and using the result to eliminate Et rt+1 yields



1
rt =
ift Et t+1
1+D


 

  D 
D
1
f
+
Et it+1 Et t+2 +
Et rt+2
1+D
1+D
1+D
Continuing to recursively substitute forward results in
i 


1
D
rt =
Et ift+i t+1+i
1 + D i=0 1 + D
under the assumption that limi

D
1+D

i

Et rt+i = 0.

6. Ball (1997) uses the following two equation model:


yt+1 = a1 yt a2 rt + ut+1
t+1 = t + yt + t+1
The disturbances ut and t are taken to be serially uncorrelated. At time t,
the policy maker chooses rt , and the state variable at time t is t +yt t .
Assume the policy makers loss function is given by equation (10.54). The
optimal policy rule takes the form t = At where t a1 yt a2 rt . Derive
the optimal value of A.
Note: In the text, the coefficient on rt in the definition of t is incorrectly
labelled as a3 .
First rewrite the model in terms of t and t :
yt+1 = t + ut+1
t+1 = t + t+1
Though its choice of rt , the policy maker can determine t , so it simplifies the
problem to simply treat t as the policy instrument.
The loss function (10.54) can now be written as

2 

1 i
L = Et
(t + ut+1 )2 + t + t+1
2 i=1

The objective is to minimize this subject to the constraint that


t+1

= t+1 + yt+1
= t + t + t+1 + ut+1
106

Define the value function




2


1 
1
2
Et (t + ut+1 ) + Et t + t+1 + Et V t + t + t+1 + ut+1
V (t ) = min
t
2
2
Then the first order conditions include


t + Et V  t + t + t+1 + ut+1 = 0


V  (t ) = t + Et V  t + t + t+1 + ut+1

(158)
(159)

Multiplying the second of these by and adding it to the first yields


t + V  (t ) = t
or

 

V (t ) = t
t

This implies
 

Et t+1

 

= t + t
Et t+1

Et V  (t+1 ) = Et t+1

Substituting this back into (158),



 

t + t + t
Et t+1 = 0

or

t =
+ 2


t +

+ 2


Et t+1

When policy is set at time t, t summaries the state, so optimal policy, given
the linear-quadratic structure, will be of the form t = At . Using this proposed
policy t = At , and recalling that Et t+1 = AEt t+1 = A(1 + A)t , this
becomes





At =
t +
A(1 + A)t
+ 2
+ 2
which yields the following quadratic equation for A:


A2 + 2 A + = 0
107

the solutions of which are


A1 =



+ 2 + ( + 2 )2 + 4 2 2
2

and
A2 =



+ 2 ( + 2 )2 + 4 2 2
2

To determine which of these solutions we want, note that


t+1 = t + t = (1 + A) t
so that t will be stable only if A < 0 so that the coefficient 1 + A is less than
1. Now consider the product of the two solutions A1 and A2 :

2 
2

+ 2 + 2 + 4 2 2
A1 A2 =
(2)2
=

4 2 2
1
2 2 = < 0
2
4

so one solution must be positive, the other negative. We are looking for the
negative solution, which is A2 , so our optimal policy rule is



+ 2 ( + 2 )2 + 4 2 2
t
t =
2
In terms of the interest rate actually set by the policy maker, we can use the
definition of t as a1 yt a2 rt and t as t + yt to obtain
rt

a1 yt t
a2
 
 
a1
1
=
yt
A2 (t + yt )
a2
a2
=

which is in the form of a Taylor rule:



 

a1 A2
A2
rt =
yt
t
a2
a2

108

Typos

The most up-to-date list of known typos can be found through my web page at
http://econ.ucsc.edu/~walshc/ or by going directly to http://econ.ucsc.edu/~walshc/typos.html.
Chapter 8
1. Page 382, Problem 1: The coefficient on l should be , not a.
2. Page 384, Problem 8: The central banks loss function should be the
discounted sum of the single period loss function given by equation (8.2).
3. Page 384, Problem 11: The weights on output and inflation in the loss
function should be ( ) and (1 + ) so that they sum to 1 regardless of
the realization of . Also set a = 1.
Chapter 9
1. Page 429, Problem 4: The term involving output in the loss function
should be the output gap, y yn , rather than simply y. Alternatively, one
could just set yn = 0 as a normalization.
Chapter 10
1. Page 452, unnumbered equation at bottom of page: The expression for the
impact of a money supply shock on the long term nominal interest rate
should be multiplied by 1/2 (see the coefficient on m in equation 10.36).
2. Page 476, Problem 6. In the last line of the problem, the coefficient on r
in the definition of should be a2 , not a3 as appears.

109

Anda mungkin juga menyukai