Chapter 5

The Basic Monetary Model

Carlos A. Végh
University of Maryland and NBER
Draft: January 2011

1 Introduction
Up to this point, we have only looked at real models that have completely
abstracted from monetary considerations. The second part of this book –which
comprises Chapter 5 through 9 –looks at the foundations of monetary economics
in an open economy. This chapter starts our journey into this monetary world
by introducing money into the endowment economy of Chapter 1. To focus
exclusively on monetary phenomena (as opposed to focusing on the interaction
between monetary and real phenomena), we introduce money in such a way that
it acts as a “veil”in the sense that the economy’s real variables (i.e., consumption
and the external accounts) are independent of the path of monetary variables
such as the money supply and the exchange rate.
In this context, we look at some basic monetary experiments that will enable
us to understand monetary considerations in isolation from the rest of the econ-
omy. Section 2 starts by de…ning the two basic regimes under which an open
economy can operate: predetermined exchange rates and ‡exible exchange rates.
The fundamental di¤erence between these two regimes is that while under ‡ex-
ible exchange rates the monetary authority controls the path of the nominal
money supply, under predetermined exchange rates the nominal money supply
is endogenously determined. Section 3 then shows that monetary and exchange
rate policy are both neutral (changes in the level of the money supply or the
exchange rate do not a¤ect the real sector) and superneutral (changes in the
rate of growth of the money supply or the exchange rate do not a¤ect the real
sector). Section 4 then proceeds to show that there is a fundamental equivalence
between predetermined and ‡exible exchange rates. Speci…cally, for any given
Comments very welcome. This chapter is part of a graduate textbook on “Open Economy
Macroeconomics in Developing Countries”, currently under preparation by the author and
should be cited accordingly. I am extremely grateful to Daniel Hernaiz, Pablo Lopez Murphy,
Rong Qiang, Belen Sbrancia, and Igor Zuccardi for their help in the preparation of this chapter.
I am also grateful to Ed Bu¢ e, Guillermo Vuletin, Yossi Yakhim, and many of my students
at UCLA and Maryland for helpful comments and suggestions.

1
path of the nominal exchange rate set by the monetary authority under prede-
termined exchange rates, there will be some endogenously-determined path of
the nominal money supply. If, under ‡exible exchange rates, the monetary au-
thority set this precise path of the nominal money supply, the same equilibrium
would obviously obtain.
While this fundamental equivalence is an important conceptual benchmark,
it should not be taken to imply that in the real world exchange rate regimes are
irrelevant. Consider, for instance, an economy operating under predetermined
exchange rates. Given that the economy will be subject to a myriad of monetary
and real shocks, the resulting equilibrium path of the nominal money supply
will exhibit high variability. Such a highly variable path would not be a path
that would be set by the monetary authority under ‡exible rates. Rather,
under ‡exible rates, the monetary authority would choose a more stable and
predictable path since, after all, the whole idea of choosing a nominal anchor is
to provide a stable nominal foundation to the economy. Put di¤erently, since, in
practice, the monetary authority will choose relatively simple paths of either the
nominal exchange rate or the money supply, the economy will react di¤erently
to the same shocks.
We illustrate this idea in Section 5 by analyzing the di¤erent response of the
economy to money shocks depending on the exchange rate regime. Suppose that
there is a negative money demand shock that will occur with certainty sometime
in the future (say at time T ). Under predetermined exchange rates, real money
balances will only fall at the moment that the shocks hits. This fall will come
about by the public buying foreign bonds from the Central Bank in exchange
for domestic money. In sharp contrast, under ‡exible exchange rates the private
sector as a whole has no means of getting rid of unwanted money balances at
time T . Further, the nominal exchange rate cannot jump at time T because, if
it did, there would be unbounded pro…t opportunities. As a result, real money
balances must begin to fall in anticipation of the negative shock, which requires
high in‡ation before the shock. The anticipation of the negative money demand
shock will have in‡ationary consequences under ‡exible exchange rates but not
under predetermined exchange rates.
Until this point in the chapter, money has been introduced by entering real
money balances as an argument in the utility function. While this is a convenient
shortcut, it does not make explicit the trading environment that may underlie
the presence of money. A more explicit framework is analyzed in Section 6
where we switch from continuous to discrete time and introduce money via a
cash-in-advance constraint. We adopt a formulation – originally due to Lucas
(1982) –in which asset markets open before goods markets, which implies that
money continues to be a veil.1
In sum, this chapter focuses on the simplest monetary model of a small
open economy in which money is a veil in the sense that monetary/exchange
1 As will become clear in Chapter 7, alternative timing assumptions in a discrete time

environment (i.e., goods markets open before asset markets) or a continuous-time formulation
of a cash-in-advance model imply that temporary changes in monetary/exchange rate policy
will a¤ect the real sector.

2
rate policy does not a¤ect the real sector. Subsequent chapters will introduce
various frictions into this benchmark model that will “remove the veil” and
allow monetary variables to a¤ect the real sector.

2 The basic monetary model

Consider a small open economy inhabited by a large number of identical, in…nitely-
lived consumers who are endowed with perfect foresight. The economy is per-
fectly integrated with the rest of the world in both goods and capital markets.
There is only one (tradable and non-storable) good, whose price is given by the
law of one price. The economy receives a ‡ow endowment of the good (yt ). The
international real interest rate (r) is given and constant over time.

2.1 Consumer’s problem

2.1.1 Budget constraints
The consumer holds two assets: domestic money (M ) and an internationally-
traded bond denominated in the foreign currency (B ). Nominal asset holdings
are therefore:

At = Mt + Et Bt , (1)
where E is the nominal exchange rate (units of domestic currency per unit of
foreign currency). By the law of one price,

Pt = Et Pt ; (2)
where Pt is the domestic currency price of the good and Pt is the foreign
currency price. By di¤erentiating equation (2) with respect to time, we obtain

t = "t + t, (3)
where t ( P_t =Pt ) is the rate of in‡ation, "t ( E_ t =Et ) is the rate of change of
the exchange rate, and t ( P_t =Pt ) is the foreign in‡ation rate.2
The numeraire of this economy will be the tradable good. Hence, “real”
variables will be de…ned in terms of tradable goods. Dividing (1) by Pt , we
obtain

at = mt + bt , (4)
where at ( At =Pt ), mt ( Mt =Pt ), and bt ( Bt =Pt ) denote real …nancial assets,
real money balances, and real foreign bonds, respectively.
The consumer’s ‡ow constraint in nominal terms is given by

A_ t = Et it Bt + E_ t Bt + Pt yt + Pt t Pt ct ; (5)
2 As a matter of terminology, we will refer to " as the rate of devaluation under predeter-

mined exchange rates and as the rate of depreciation under ‡exible exchange rates.

3
where it denotes the foreign nominal interest rate, yt is the endowment of the
good, t denotes real lump-sum transfers from the government and ct denotes
consumption. The term Et it Bt captures interest payments on the foreign bonds
(in terms of domestic currency), while the term E_ t Bt denotes capital gains on
the foreign bonds:
To express the ‡ow constraint in real terms, divide (5) by Pt (taking into
account the law of on price) to obtain

A_ t
= (it + "t )bt + yt + t ct ; (6)
Pt
_
where "( E=E) denotes the rate of depreciation/devaluation. Given that, by
de…nition, at = At =Et Pt , it follows that

A_ t
a_ t = ("t + t )at . (7)
Pt
Substituting (6) into (7) and rearranging terms:

a_ t = (it t )at + yt + t ct (it + "t )mt : (8)

Assuming that the Fisher equation holds in the rest of the world (i.e., it = r+ t )
and taking into account that perfect capital mobility implies that interest parity
will hold (i.e., it = it + "t ), we can rewrite (8) as

a_ t = rat + yt + t ct it mt : (9)
Integrating forward equation (9) and imposing the transversality condition
lim at e rt = 0 (for the reasons discussed in Chapter 1), we …nally obtain
t!1
Z 1 Z 1
a0 + (yt + t )e rt dt = (ct + it mt )e rt dt: (10)
0 0
This lifetime constraint makes perfect sense as it says that the present dis-
counted value of “total expenditures”(given by the RHS, and which include the
opportunity cost of holding real money balances) must equal the consumer’s
wealth (LHS), which comprises his/her initial real …nancial assets (a0 ) and the
present discounted value of the endowment and government transfers.

2.1.2 Utility maximization

The consumer’s lifetime utility is given by
Z 1
t
[u(ct ) + v(mt )]e dt; (11)
0

where (> 0) is the subjective discount rate, and the functions u(ct ) and v(mt )
are strictly increasing and strictly concave in their arguments:

4
 u0 (ct ) > 0; v 0 (mt ) > 0;
u00 (ct ) < 0; v 00 (mt ) < 0:

The rationale behind introducing money in the utility function is that real
money balances provide liquidity services that can be thought of as propor-
tional to the stock of real money balances. In other words, liquidity services
are captured by mt where, for simplicity, it is assumed that = 1. Hence, the
way to think of v(m) is that consumers derive utility from the liquidity services
provided by money.3
The consumer’s problem consists in choosing fct ; mt g for all t 2 [0; 1) to
maximize lifetime utility (11) subject to the lifetime constraint (10), for a given
path of t , it , and yt and given values of r and a0 .
Assuming, as usual, that = r, the …rst order conditions imply

u0 (ct ) = ; (12)
v 0 (mt ) = it : (13)

Combining these two equations, we obtain

v 0 (mt )
u0 (ct ) = ;
it
which implicitly de…nes a real money demand with standard properties:

mt = L(ct ; it ); (14)
@L it u00 (ct )
= > 0; (15)
@ct v 00 (mt )
@L u0 (ct )
= < 0: (16)
@it v 00 (mt )

Real money demand is thus increasing in consumption and decreasing in the

nominal interest rate (the opportunity cost of holding money).4

2.2 Government
The government comprises the …scal authority and the monetary authority (i.e.,
the Central Bank). Let Ht be the amount of net foreign bonds (measured in
terms of foreign currency) that the government holds and Ht ( Et Ht ) denote
3 There are two other popular ways of introducing money: via a cash-in-advance constraint

(as analyzed later in this chapter and in Chapter 7) or a transactions costs technology (Chapter
7). All three ways have pros and cons that will be made clear as we proceed further.
4 It is worth noting that, in any microfounded model of money, real money demand will

depend on consumption, as opposed to income or output.

5
the domestic currency value of these bonds. The government’s ‡ow constraint
in nominal terms is given by5
:
H_ t = it Et Ht + E_ t Ht + M_ t Pt t . (17)
| {z } | {z } |{z} |{z}
interest incom e capital gains m oney printing transfers

As indicated below the equation, the government has three sources of revenues:
(i) interest income on its international reserves; (ii) capital gains on its interna-
tional reserves; and (iii) money printing or revenues from money creation. The
only government expenditure consists of lump-sum transfers.6
To obtain the government’s ‡ow constraint in real terms, we proceed in the
same way as we did above for the consumer. De…ne the real value of international
reserves (i.e., international reserves in terms of real dollars) as h( H=P ). It
then follows that

H_ t
h_ t = ("t + t )ht . (18)
Pt
Dividing (17) by Pt , substituting the resulting expression into (18) (and impos-
ing the Fisher equation for the rest of the word), we obtain:
:
Mt
h_ t = rht + t. (19)
Pt
It proves illuminating to write the real revenues from money creation, Mt =Pt ,
as
:
Mt
=m
_ t + ("t + t )mt : (20)
Pt
As indicated, real revenues from money creation can be divided into two com-
ponents: (i) seigniorage (…rst term on the RHS), which refers to the revenues
that may accrue to the government as the result of an increase in the public’s
real demand for money; and (ii) the in‡ation tax (second term on the RHS),
which refers to the revenues that may accrue to the government as a result of
the public’s desire to replace the loss of value of real money balances due to a
positive in‡ation rate.
Substituting equation (20) into equation (19), we obtain

h_ t = rht + m
_ t + ("t + t )mt t. (21)
5 Appendix 7.1 breaks down the government into the monetary and the …scal authority

and shows how equation (17) follows from consolidating their respective nominal budget con-
straints.
6 At this point, we are, of course, abstracting from tax revenues from conventional taxes

(i.e., taxes other than the in‡ation tax) and government spending.

6
 Integrating forward equation (19) and imposing the transversality condition
lim ht e rt = 0, we obtain the government’s intertemporal constraint:7
t!1
Z 1
: Z 1
Mt rt rt
h0 + e dt = te dt. (22)
0 Pt 0
In closing, notice that so far we have only looked at …scal accounting and
made no policy assumptions.

2.3 Equilibrium conditions

The assumption of perfect capital mobility implies that interest parity holds:

it = it + "t . (23)
Let kt ( bt +ht ) denote the economy’s stock of net foreign assets. Combining
the consumer’s ‡ow constraint (equation (9)) with the government’s (equation
(21)) yields the economy’s ‡ow constraint:

k_ t = rkt + yt ct . (24)
The economy accumulates net foreign assets (i.e., k_ t > 0) to the extent that
the economy’s income (rkt + yt ) exceeds the economy’s consumption (ct ). To
link the economy’s ‡ow constraint to standard balance of payments accounting,
rewrite this equation as

h_ t = b_ t + r(bt + ht ) + yt ct , (25)
|{z} |{z} | {z } | {z }
h KA IB TB
| {z }
CA

where TB, IB, CA, KA, and h denote trade balance, income balance, current
account, capital account, and increase in international reserves. As indicated,
therefore, equation (25) constitutes the fundamental identity of balance of pay-
ment accounting which, in words, reads as8

Increase in international reserves = Capital account + Current account.

In the “real” world of Chapter 1, a current account de…cit had to be …nanced

by a capital account surplus (i.e., by borrowing from abroad). In contrast,
a monetary economy can run a current account de…cit and a capital account
de…cit (i.e., lending abroad) provided that the Central Bank is …nancing these
two de…cits by losing international reserves.
7 Strictly speaking, the RHS of constraint (22) should include all future jumps in real money

balances. For example, if such a jump occurred at time T , then we would include a term of
the form e rT m.
8 As a matter of terminology, notice that the change in international reserves is often referred

to as the “balance of payments”. If the country is gaining (losing) international reserves, the
balance of payments is positive (negative).

7
 Finally, notice that integrating forward the economy’s ‡ow constraint (equa-
tion (24)) and imposing the corresponding transversality condition yields the
economy’s resource constraint:
Z 1 Z 1
k0 + yt e rt dt = ct e rt dt. (26)
0 0

2.4 Perfect foresight equilibrium

We now characterize the perfect foresight equilibrium paths (PFEP) of con-
sumption and real money balances. It follows from …rst-order condition (12)
that consumption will be constant along a perfect foresight equilibrium. Then,
from the economy’s resource constraint (26),
Z 1
c = r k0 + yt e rt dt . (27)
0

Due to the separability of consumption and real money balances in the utility
function, consumption is constant over time (and equal to permanent income)
regardless of the path of the nominal variables. As in the endowment model of
Chapter 1, consumption will be ‡at over time for any path of the endowment. In
periods of low (high) endowment, the economy will run trade de…cits (surpluses)
to smooth consumption over time. Furthermore, consumption and the external
accounts would respond to unanticipated endowment shocks in exactly the same
way as they did in Chapter 1. Hence, in this basic monetary model, money is
just a “veil” in the sense that the model exhibits the “dichotomy” between the
real and monetary sectors emphasized in classical monetary economics.9
From (14), real money demand along a PFEP will be given by

mt = L(c; it ). (28)
Hence, real money demand will be constant if the nominal interest rate is
constant over time. We now turn to the determination of the nominal interest
rate and other nominal variables. (In what follows we will assume that the
foreign in‡ation rate is constant at the level .)

2.5 Nominal anchors

The determination of the nominal interest rate and the paths of both the nom-
inal exchange rate and the nominal money supply will depend on the speci…c
monetary regime adopted by the monetary authority. We will now study the de-
termination of these variables under the two main nominal anchors: the exchange
rate (predetermined exchange rates) and the money supply (‡exible exchange
rate).10
9 It is worth noticing that this dichotomy refers to the fact that the monetary sector does

not in‡uence the real sector. But real shocks, of course, may a¤ect monetary variables.
1 0 In most modern countries, the central bank has a legal monopoly over the issuance of

money. As with any monopolist, the central bank can therefore either set the price (and let

8
 As a starting point, consider the Central Bank’s balance sheet illustrated in
Table 1.11

[Table 1]
From the balance sheet, it follows that

Et Ht + Dt = Mt , (29)
12
where Dt denotes the stock of nominal domestic credit.
To organize our thoughts regarding the mechanics of nominal anchors and
the determination of nominal magnitudes (we will deal with the economics be-
low), it is helpful to rewrite equation (28) as
Mt
= L(c; it ); (30)
Et Pt
where Pt is, of course, exogenously-given. As will be clear below, in a station-
ary equilibrium, the nominal interest rate, it , will be determined by the rate of
growth of the nominal anchor. Hence, let us think of it as having been deter-
mined already for the purposes of thinking about the determination of nominal
magnitudes. We can then viewed equations (29) and (30) as a two-equation
system with four unknowns: Et , Ht , Dt , and Mt . The monetary authority
can then set two of these four variables and let the other two be determined
endogenously. In addition, equation (30) tells us that either Mt or Et must be
set by the monetary authority (but not both because, in that case, equation
(30) would be overdetermined).
In this context –and as Table 2 illustrates –we can de…ne a predetermined
exchange rate regime as one in which the monetary authority sets Et and Dt and
lets Mt and Ht be determined endogenously. Put di¤erently, we can think of
the monetary authority as setting Et which, through equation (30), determines
Mt . With Et and Mt so determined and Dt being set by the monetary authority,
the Central Bank’s balance sheet (29) endogenously determines Ht .

[Table 2]
the quantity be market-determined) or the quantity (and let the price be market-determined).
The former is the case of predetermined exchange rates and the latter is the case of ‡exible
exchange rates.
1 1 Following common practice – and for simplicity – the balance sheet assumes that the

central bank’s capital is zero.

1 2 In the context of this model – in which we do not explicitly model domestic bonds –

the label “domestic credit” is somewhat of a misnomer. The label “domestic credit” is more
natural when the monetary authority introduces money into the system via open-market
operations whereby a purchase of domestic bonds (an increase in domestic credit) increases
the money supply and a sale of domestic bonds (a reduction in domestic credit) reduces
the money supply. In the current model, however, money should be thought of as being
introduced into the economy through a “helicopter drop” (and being retired from the system
with a giant vacuum cleaner!). Hence, “domestic credit” is essentially an accounting …ction
that could, however, be interpreted as a non-interest bearing obligation of the private sector
vis-a-vis the monetary authority.

9
 To de…ne a ‡exible exchange rate regime, we need to add an additional
wrinkle. Under ‡exible exchange rates, the nominal exchange rate will be an
endogenous variable. The Central Bank’s balance sheet (29) would then tell
us that, for given Ht and Dt , changes in E would change the money supply.
For instance, an increase in the nominal exchange rate would lead the Central
Bank to print more money. In practice, however, Central Banks typically do
not “monetize” changes in the nominal exchange rate and simply credit/debit
a non-monetary liability. To …x ideas, denote this non-monetary liability by
N M and rewrite the balance sheet of the Central Bank as

Et Ht + Dt = Mt + N Mt . (31)
We can then view equations (30) and (31) as a system of two equations in 5
unknowns: Et , Ht , Dt , Mt , and N Mt . The Central Bank can then set three
of these …ve variables and let the other two be determined endogenously (see
Table 2) . Under ‡exible exchange rates, the Central Bank sets D, M , and
Ht and lets Et and N Mt be determined endogenously. In other words, we can
think of the Central Bank setting Mt and the money market equilibrium (30)
determining Et . With Mt and Et so determined and the Central Bank setting
H , equation (31) determines N Mt .
Having discussed the mechanics behind predetermined and ‡exible exchange
rate regimes, we now turn to a more detailed discussion and solve the model
under both regimes.

2.5.1 Predetermined exchange rates

Under predetermined exchange rates, the monetary authority sets the path of
the nominal exchange rate (Et ) and the path of the nominal stock of domestic
credit (Dt ). The path of international reserves (Ht ) and the path of the nominal
supply of money (Mt ) will be endogenously determined. The idea that the
nominal money supply is endogenous under predetermined exchange rates is
critical to the understanding of this regime.
We use the term “predetermined exchange rates”rather than the more com-
mon label “…xed exchange rates” because, strictly speaking, the latter is just
a particular case of the former. The key feature of a predetermined exchange
rate regime is that, at any given point in time, the monetary authority stands
ready to buy and sell foreign exchange at a given price (i.e., at a given exchange
rate). In other words, the public can go to the Central Bank and sell foreign
exchange (e.g., dollars) to the Central Bank in exchange for domestic currency
and viceversa. If the price at which the Central Bank sells/buys foreign ex-
change is constant over time, we then talk of a …xed exchange rate regime.
But if that price varies over time, then we need to use the more general term
“predetermined exchange rate”.13
Formally, setting the path of the nominal exchange rate implies setting the
initial level, E0 , and a constant rate of growth, ", of the nominal exchange
1 3 Box 1 lists di¤erent kinds of predetermined exchange rate regimes.

10
rate.14 Given ", the interest _parity condition (23) determines a constant level
of the nominal interest rate, i :
_
i = i + ". (32)
The constancy of the nominal interest rate implies, from (28), that real
money balances will be constant over time at a level given by:
_
m = L(c; i ) (33)
Hence, m_ t = 0 for all t 2 [0; 1). Since, by de…nition, m = M=EP , it
follows that the rate of money growth will also be constant over time:

="+ . (34)
The only nominal variable yet to be determined is the initial level of the
nominal money supply, M0 . Since equation (28) holds at time 0, we can write:
M0 _
= L(c; i ):
P0 E 0
Solving for M0 :
_
M0 = P0 E0 L(c; i ): (35)
Interestingly enough, so far we have not made use of the path of nominal
domestic credit. Hence, all the endogenous variables derived above are inde-
pendent of the path of nominal domestic credit. The path of nominal domestic
credit will matter though in determining the path of international reserves (h).
As indicated above, the monetary authority sets the initial level, D0 , and the
(constant) rate of growth ( ) of domestic credit. To determine the resulting
path of international reserves, …rst express the Central Bank’s balance sheet in
real terms:
h t + d t = mt .
Then di¤erentiate this identity and solve for h_ t :

h_ t = m
_t d_t : (36)
This equation says that, under predetermined exchange rates, any creation in
real domestic credit in excess of changes in real money demand will result in
a loss of international reserves. In other words, if d_t > m _ t , then h_ t < 0.
Intuitively, money printed by the Central Bank that is not demanded by the
private sector will result in a loss of international reserves as the private sector
gets rid of unwanted money balances by exchanging them for net foreign assets at
the Central Bank. To gains additional insights, notice that, since d D=EP ,
1 4 Of course, the monetary authority could set a non-constant path of the devaluation rate.

We will study some of those cases in later chapters.

11
 d_t = dt ( " ). (37)
Substituting (37) into (36), we obtain

h_ t = m
_t dt ( " ):
In this particular case, m
_ t = 0 because, from (33), real money demand is con-
stant. Hence, we can rewrite this equation as

h_ t = dt ( " ): (38)
Given that real money demand is constant, if nominal domestic credit is growing
faster than domestic in‡ation (i.e., if > " + ), then the monetary authority
will be losing international reserves (i.e., h_ t < 0). Needless to say, this has
been a common situation in developing countries where the monetary authority
is often coerced into printing money to …nance the …scal authority’s spending.
If this situation persists and there is some threshold below which international
reserve cannot fall, the monetary authority will run out of international reserves
and a balance of payment crisis will ensue (as analyzed in detail in Chapter 9).
Conversely, a situation in which < " + would imply that the monetary
authority accumulates international reserves without bound and can also be
ruled out. Hence, for a predetermined exchange rate regime to be sustainable
over time, the rate of domestic credit growth must equal the domestic in‡ation
rate (i.e., = "+ ).15 This will be, therefore, our maintained assumption.
Having established that the path of international reserves is constant over
time, we need to establish their initial value, h0 . From the Central Bank’s
balance sheet at t = 0 and (33), it follows that
_
h0 = L(c; i ) d0 . (39)
In principle, h0 could take any sign, since the Central Bank could have a negative
asset position.
Finally, notice that the variable that adjusts to make the government con-
straint hold at any point in time is the level of transfers. From (19), and taking
into account that ht , "t , and mt are all constant over time, it follows that

t = rh + (" + )m. (40)

To …x ideas, consider a …xed exchange rate regime (" = 0) and zero foreign
in‡ation. Then, t = rh. If h < 0, then the government is …nancing the debt
service by taxing the private sector.
1 5 We should note that, strictly speaking, this is a su¢ cient, though not necessary, condition

for a predetermined exchange rate regime to be sustainable (see Appendix 7.3).

12
2.5.2 Flexible exchange rates
Under ‡exible exchange rates, the monetary authority does not intervene in
the foreign exchange market and allows the exchange rate to be determined by
market forces. If the Central Bank does not intervene in the foreign exchange
market, its international reserves will be constant over time.16 For simplicity,
it is typically assumed that this initial level of international reserves is zero.
If international reserves are zero, then the Central Bank’s balance sheet
reduces to:

Dt = Mt .
Hence, under ‡exible exchange rates, setting the path of nominal domestic credit
is equivalent to setting the path of the nominal money supply.17 In particular,
the monetary authority sets the initial level, M0 , and a constant rate of growth,
, of the nominal money supply.
We …rst show that real money balances will be constant over time. To see
this, notice that m_ t =mt = "t , and use (13) and (23) to obtain:

v 0 (mt )
m
_ t = mt r + : (41)

Linearizing this equation around the stationary value for real money balances
(given by r + = v 0 (mt )= ), we see that this is an unstable di¤erential equa-
tion.18 Formally,
@m_t mss v 00 (mss )
= > 0.
@mt ss
This implies that unless m is already at its stationary value at t = 0, it will
diverge over time. Intuitively, if m increases the nominal interest rate must
fall to accommodate this increase. By interest parity, this implies a fall in the
rate of depreciation (in‡ation), which in turn implies that real money supply
will grow faster which requires a further fall in the nominal interest rate and so
forth. Hence, the only convergent equilibrium path is for real money balances
to be constant and (implicitly) given by

r+ = v 0 (m)= ;
for all t 0. Given this value for_ real money balances, equation (28) determines
a unique nominal interest rate, i . Further, since m
_ t =mt = "t = 0, the
(constant) rate of depreciation will be given by:
1 6 The central bank could, of course, choose a non-constant path of international reserves

by appropriately intervening in the foreign exchange market. This would be the case of “dirty
‡oating” analyzed in Exercise 1 at the end of this chapter.
1 7 If the constant level of reserves had not been set equal to zero, then this statement would

not be quite correct because changes in the nominal exchange rate would imply capital gains
or losses, which would a¤ect the nominal money supply. In practice, however, central banks
do not “monetize” capital gains or losses but instead debit a non-monetary liability.
1 8 As an example, notice that if v (m) = log(m), then equation (41) becomes a linear di¤er-

ential equation given by m _ t = (r + )mt 1= which is, of course, unstable.

13
 "= . (42)
From the interest parity condition (23) and (42) – and taking into account
that i = r + –it follows that the constant level of the nominal interest rate
is given by:
_
i =r+ .
The only nominal variable yet to be determined is the initial level of the
nominal exchange rate, E0 (i.e., the initial price level). Since equation (28)
holds at time 0, we can write:

M0 _
= L(c; i ):
P0 E 0
Solving for E0 :

M0
E0 = _ :
P0 L(c; i )
We have thus shown that, as under predetermined exchange rates, the value of
all nominal variables is perfectly well-de…ned under ‡exible exchange rates.
Finally, how does the level of transfers get determined? Since international
reserves are equal to zero and m _ t = 0, then from (21):

t = (" + )m. (43)

3 Neutrality and superneutrality results

We now proceed to show that, in our basic monetary model, monetary and
exchange rate policy are neutral and superneutral. By “neutral”monetary pol-
icy, we mean that an (unanticipated) and permanent change in the level of the
money supply has no real e¤ects.19 It simply leads to an equi-proportional
change in the exchange rate. By the same token, a neutral exchange rate policy
means that a permanent devaluation has no real e¤ects and leads to an equipro-
portional change in the nominal money supply. By “superneutral”monetary or
exchange rate policy, we mean that neither a change in the rate of change of
the money supply or in the rate of devaluation has any real e¤ects.
Since both monetary and exchange rate policy are neutral and superneutral,
changes in monetary/exchange rate policy have no e¤ects on the real economy.
Conversely, real shocks will have the same real e¤ects under either ‡exible or
predetermined exchange rate regimes, as examined in Exercise 2 at the end of
the chapter. Hence, money is a veil in this basic model in the sense that there
is no interaction between real and monetary variables.
1 9 In this context, it is always implicitly understood that “real e¤ects”refers to real variables

other than real monetary balances (which may change).

14
3.1 Exchange rate policy
We will now study the e¤ects of (i) an unanticipated and permanent devaluation
(i.e., an increase in the level of the exchange rate), and (ii) an unanticipated
and permanent increase in the devaluation rate.

3.1.1 A permanent devaluation

Suppose that the economy is initially in the stationary perfect foresight equi-
librium characterized above (for = 0). For simplicity, assume that initially
" = = 0, so that the exchange rate is initially …xed. At t = 0, there is an
unanticipated and permanent increase in the level of the nominal exchange rate
(i.e., a permanent devaluation); see Figure 1, Panel A. What will be the e¤ects
of this devaluation?

[Figure 1]
Clearly, the devaluation has no real e¤ects since we have already shown
that, along a PFEP, consumption is given by (27) regardless of the path of the
exchange rate (Figure 1, Panel B).
From (32), we also see that the nominal interest rate will not change (Fig-
ure 1, Panel C). Hence, from (33), real money demand does not change either
(Figure 1, Panel D). From (34), the same is true of the rate of money growth.
From (35), we see that the initial level of nominal money balance will increase
in the same proportion as the exchange rate.
The main action resulting from a permanent devaluation actually takes place
in the Central Bank’s balance sheet. Since the stock of nominal domestic credit
is controlled by policymakers and hence is given at t = 0, the real stock of
domestic credit falls at t = 0 and remains at that lower level thereafter (Figure
1, Panel E). We can then infer the path of international reserves from the Central
Bank balance sheet (ht = m dt ). It follows that international reserves jump
up on impact and remain constant thereafter (Figure 1, Panel F). In fact –
as follows from the Central Bank’s balance sheet – the change in international
reserves at t = 0 is exactly equal to the reduction in the real stock of domestic
credit:

h0 = d0 > 0:
We thus conclude that a devaluation leads to a gain in international reserves.
What is the intuition behind the increase in international reserves at the
Central Bank? The key is that while the devaluation does not a¤ect real money
demand, it reduces real money supply for the initial nominal money supply. In
other words, at the initial money supply, there is an incipient excess demand
for money. To rebuild their money balances, consumers go to the Central Bank
and exchange net foreign assets for nominal money balances.20
2 0 An alternative interpretation, which focuses on Central Bank intervention, goes as follows.

To get rebuild money balances, consumers sell foreign assets. This puts downward pressure

15
 The idea of an “excess money demand” leading to a gain in international
reserves –which occurs instantaneously in this model with no frictions –is one of
the most fundamental monetary adjustment mechanisms under predetermined
exchange rates. Introducing various frictions into this basic model (such as
assuming that there are no interest bearing bonds as in Chapter 6 or assuming
that there are capital controls) will force this adjustment to take place gradually
over time but will not alter its fundamental nature.21

3.1.2 A permanent increase in the rate of devaluation

Suppose now that, starting from the same initial equilibrium (for = " =
= 0), there is an unanticipated and permanent increase in the devaluation
rate (Figure 2, Panel A).22 Once again, consumption remains constant (Figure
2, Panel B). From (32), we infer that the nominal interest rate increases pari
passu with the devaluation rate (Figure 2, Panel C). Since the opportunity cost
of holding money increases, real money demand falls at t = 0, as follows from
(33) (Figure 2, Panel D). The rate of money growth increases (from (34)). From
(35), the initial level of the money supply also falls. The path of real domestic
credit remains unchanged (Figure 2, Panel E). Finally, since real money demand
falls, we infer from (39) that international reserves fall at t = 0.

[Figure 2]
How does this adjustment take place? In response to the increase in the
opportunity cost of holding money, the consumer wants to reduce his/her real
money holdings. To do so, he/she goes to the Central Bank and exchanges
domestic money for foreign assets (i.e., sells domestic currency and buys foreign
assets). As a result, international reserves at the Central Bank fall whereas
private holdings of net foreign assets go up. For the economy as a whole, net
foreign assets do not change.
It is interesting to note that while a permanent devaluation leads to an
increase in international reserves, a permanent increase in the devaluation rate
results in a fall in international reserves. The di¤erence is due to the di¤erent
way in which equilibrium in the money market is a¤ected. In the …rst case –
and for the initial nominal money supply –the devaluation leads to an incipient
excess real money demand which requires an upward adjustment in the nominal
on the nominal exchange rate (the nominal price of foreign bonds). To prevent the domestic
currency from appreciating, the central bank must step in and buy the foreign assets o¤ered
by the private sector. By so doing, the central bank increases the money supply until the
money market is in equilibrium, at which point there are no further pressures on the nominal
exchange rate.
2 1 By the same token, a revaluation (i.e., a fall in E) would lead to a loss in international

reserves because, at the initial nominal money supply, there would be an excess supply of
money.
2 2 To ensure that the predetermined exchange rate regime continues to be sustainable, we

assume that increases by the same amount.

16
money supply.23 In the second case –and for the initial nominal money supply
–the increase in the rate of devaluation leads to an incipient excess real money
supply which requires a fall in the nominal money supply.

3.2 Monetary policy

We now turn to ‡exible exchange rates and examine the e¤ects of (i) an unan-
ticipated and permanent increase in the stock of nominal money and (ii) an
unanticipated and permanent increase in the rate of money growth.

Permanent increase in money supply Suppose that, starting from the

initial equilibrium described above (with = = 0), there is an unantici-
pated and permanent increase in the nominal money supply (Figure 3, Panel
A). Consumption, of course, remains constant (Figure 3, Panel B). In the new
equilibrium, real money balances must remain constant for, if they did not, the
path of real money balances would diverge over time (Figure 3, Panel C). The
constancy of real money balances implies that, on impact, the nominal exchange
rate will increase by the same proportion as the nominal money supply and re-
main at that level thereafter (Figure 3, Panel D). The fact that m _ t = 0 for all t
implies that the rate of depreciation continues to be equal to 0 (Figure 3, Panel
E). This implies that the nominal interest rate also remains constant (Figure 3,
Panel F).

[Figure 3]
We conclude that a permanent increase in the nominal money supply leads to
an equi-proportional increase in the nominal exchange rate, leaving unchanged
all other variables.

Permanent increase in the rate of money growth Suppose that, starting

from the initial equilibrium described above (with = = 0), there is an
unanticipated and permanent increase in the rate of money growth (Figure 4,
Panel A). Consumption, of course, remains unchanged (Figure 4, Panel B).
Since real money balances are governed by an unstable di¤erential equation,
they must adjust immediately to their new and lower stationary value (Figure
4, Panel C). Otherwise, the path of real money balances would diverge over time.
Since the nominal money supply is given at t = 0, we infer that the nominal
exchange rate increases on impact (Figure 4, Panel E). The fact that m _t =0
for all t implies that the rate of depreciation increases on impact and remains
at that level thereafter. This, in turn, implies that the nominal exchange rate
increases at the rate of money growth from t = 0 onwards (Figure 4, Panel
D). By interest parity, the nominal interest rate increases on impact and stays
constant thereafter (Figure 4, Panel F).
2 3 In contrast, a permanent increase in domestic credit (i.e., an increase in D ) would lead
t
to an incipient excess suppy of money and hence a loss of international reserves, as analyzed
in Exercise 3 at the end of the chapter.

17
 [Figure 4]
We conclude that a permanent increase in the rate of money growth leads to
an increase in the exchange rate (a nominal depreciation) and a corresponding
increase in the rate of devaluation and the nominal interest rate.

4 Equivalence results
We will now show that there is a fundamental equivalence between predeter-
mined and ‡exible exchange rates. Consider an economy operating under prede-
termined exchange rates. (For simplicity, assume foreign in‡ation is zero.) For
any given path of the exchange rate set by the monetary authority, there will be
a corresponding path of the nominal money supply. If, under ‡exible rates, the
monetary authority were to set exogenously that path of the nominal money
supply, the same equilibrium paths would obtain. In fact –and unless you were
able to see the movements in international reserves at the Central Bank – you
would be unable to tell the two regimes apart.
We will illustrate this point by looking at two examples. First, we will
assume that the economy is operating under a predetermined exchange rate
regime with a non-constant rate of devaluation and study the implications for
the path of the nominal money supply. We will then examine the opposite case:
we will assume that the economy is operating under ‡exible exchange rates with
a non-constant path of the money growth rate and examine the implications for
the path of the nominal exchange rate.24

4.1 A non-constant rate of devaluation

Suppose that the economy is operating under predetermined exchange rates.
Consider the perfect foresight equilibrium corresponding to the path of the
devaluation rate illustrated in Figure 5, Panel A. The rate of devaluation is
constant until time T (at the level "H ) at which time it falls to a lower level ("L )
and stays there afterwards. The corresponding path of the nominal exchange
rate is illustrated in Figure 5, Panel B. By the interest parity condition (equation
(23)), the nominal interest rate is high and then low (Figure 5, Panel C). From
(28), real money demand will be low until time T and then jump to a higher
level at time T (Figure 5, Panel D). (Consumption is of course constant and
independent of the path of the nominal exchange rate.) The path of the nominal
money supply (Figure 5, Panel E) follows from the path of real money balances
and that of the nominal exchange rate. Since real money balances are constant
until time T , the nominal money stock must be growing at the same rate as
the nominal exchange rate ("H ). At time T , the nominal money stock increases
discretely as the public exchanges foreign assets for domestic money at the
2 4 Exercise 4 at the end of the chapter looks at a third example (an anticipated increase in

the money supply under ‡exible rates).

18
Central Bank. After time T , the nominal money supply increases at the lower
rate of devaluation ("L ). Finally, since m
_ t = 0 for all t 0, then t = ". Hence,
is …rst high and then falls at time T (Figure 5, Panel F).

[Figure 5]

Consider now this same economy operating under ‡exible exchange rates.
Further, suppose that the path of the nominal money supply set by the Central
Bank is described by Figure 5, Panel E. In other words, the Central Bank has
announced that the nominal money supply will grow at the rate "H until time
T , increase discretely at time T , and then grow at the lower rate "L . We will
now check that – as we should expect – the corresponding paths of the rate
of depreciation, the nominal exchange rate, the nominal interest rate, and real
money balances are as illustrated in Figure 5.
Recall that, under ‡exible exchange rates, real money balances are governed
by the unstable di¤erential equation given by (41). Between 0 and T , the dynam-
ics of real money balances will be governed by the laws of motion corresponding
to the stationary equilibrium given implicitly by r + "H = v 0 (m)= . At time
T , real money balances will jump since nominal money supply goes up but the
nominal exchange rate does not change. By construction (recall Figure 5, Panel
D), the jump in nominal money at time T is exactly the increase needed to
take real money balances from their level between [0; T ) to their new and higher
level. Hence, real money balances will stay at their stationary value until time
T and at that point jump to their new value. This is, of course, the same path
described by Figure 5, Panel D. Having established that m _ t = 0 for all t 0, it
then follows that the rate of depreciation is given by the rate of money growth
at all points in time and thus follows the path illustrated in Figure 5, Panel
A. The corresponding path of the nominal exchange rate is given by Figure 5,
Panel B.
In sum, if under ‡exible rates the monetary authority sets the path of the
nominal supply given by Figure 5, Panel E, the equilibrium paths of the rate
of depreciation, the nominal exchange rate, the nominal interest rate, and real
money balances would be exactly the same as under a predetermined exchange
rates system in which the Central Bank sets the path of devaluation given
by Figure 5, Panel A. An outside observer who can only see the six variables
illustrated in Figure 5 would not be able to tell if the economy is operating under
predetermined exchange rates or ‡exible exchange rates. But, of course, if the
observer could see the Central Bank’s balance sheet, he/she would be able to tell
by the changes in the balance sheet at time T . Under predetermined exchange
rates, he/she would see the increase in the monetary base accompanied by an
increase in international reserves whereas under ‡exible exchange rates, he/she
would see an increase in nominal domestic credit.25
2 5 Of course, if under predetermined exchange rates, the central bank increased domestic

credit at T by precisely the amount needed to meet the additional real money demand, our
observer would not be able to tell which regime the economy is operating under even by
looking at the central bank’s balance sheet!

19
4.2 A non-constant path of the money growth rate
Suppose now that the economy is operating under ‡exible exchange rates and
the Central Bank sets a non-constant rate of money growth, as illustrated in
Figure 6, Panel A. The corresponding path of the nominal money supply is
depicted in Figure 6, Panel B.

[Figure 6]
To solve for the perfect foresight path of real money demand, we need to
make the critical observation that real money balances cannot jump at T . To
see this, recall that, by de…nition, m = M=EP . The path of the nominal
money supply is, by assumption, continuous at time T . The same is true of
P . Furthermore, in equilibrium, the nominal exchange rate cannot jump at T
(i.e., the nominal exchange rate cannot jump in an anticipated fashion). If it
did, there would be in…nite arbitrage opportunities which would be inconsistent
with equilibrium. For example, suppose that the exchange rate were expected
to increase at time T . Since the consumer knows this with certainty, he/she
would get rid of all money balances just an instant before the increase in the
exchange rate. Real money demand would thus fall to zero an instant before
T , which is inconsistent with equilibrium (as it would lead to an in…nite price
level or exchange rate). Conversely, if the exchange rate were expected to fall
at T , consumers would want to get rid of all net foreign assets and switch into
domestic money in anticipation of in…nite returns at time T . Money demand
just an instant before T would be in…nite which, again, is inconsistent with
equilibrium. We conclude that real money balances must be continuous at time
T.
Given that real money balances cannot jump at T , we infer that m will have
to converge in a continuous fashion to its higher stationary equilibrium. Hence,
m0 will need to be above the level corresponding to the stationary equilibrium
implicitly de…ned by r + H = v 0 (m)= so that it increases during the transition
period (Figure 6, Panel C). Since consumption is always equal to permanent
income in this model, we can infer the behavior of the nominal interest rate
from the real money demand equation (28). The nominal interest rate will thus
fall over time towards its stationary value (Figure 6, Panel D).
As for the path of the rate of depreciation, notice that
m
_t H
= "t > 0; t 2 [0; T ):
mt
We infer that during the transition the rate of depreciation will be below H .
From the interest parity condition, we also know that the rate of depreciation
will be falling over time and be continuous at T (see Figure 6, Panel E). The
corresponding path of the nominal exchange rate is depicted in Figure 6, Panel
F.
Suppose now that the economy were operating under predetermined ex-
change rates and that the Central Bank set the path for the nominal exchange
rate depicted in Figure 6, Panel F. In other words, the Central Bank would set

20
a path that would involve a declining rate of devaluation over time until it con-
verges to L . From the interest parity condition (23) , the nominal interest rate
would follow the path illustrated in Figure 6, Panel D. From the real money
demand equation and the fact that consumption is constant throughout, the
path of real money balances would be as illustrated in Figure 6, Panel C. The
change in real money balances over time combined with the rate of devaluation
would yield, by construction, the path for the rate of money growth illustrated
in Figure 6, Panel A.
Once again, if an outside observer saw the behavior of the six variables in
Figure 6, he/she would be unable to tell which exchange rate regime the economy
is operating under. The observer would need to see the Central Bank’s balance
sheet to …nd out the exchange rate regime. Under predetermined exchange
rates, the Central Bank is gaining international reserves between time 0 and
time T whereas, under ‡exible rates, domestic credit is increasing.26

4.3 On the non-equivalence in practice

Even though, as we just saw, there is a fundamental theoretical equivalence
between predetermined and ‡exible exchange rates, it does not follow that ex-
change rate regimes are irrelevant. Far from it –and as we will see in this and
subsequent chapters –the exchange rate regime is often crucial in determining
the economy’s response to various shocks. The reason is that, in practice, poli-
cymakers set relatively simple paths for the exchange rate (under predetermined
exchange rates) or for the money supply (under ‡exible exchange rates). Given
these simple paths, and the fact that economies are subject to a multitude of
both real and monetary shocks, the nominal money supply (under predeter-
mined exchange rates) or the nominal exchange rate (under ‡exible exchange
rates) will follow very volatile paths. For exchange rate regimes to be equiva-
lent in practice, we would need to observe policymakers setting, say, some simple
path for the nominal exchange rate (under predetermined exchange rates) and
then some other policymakers setting the corresponding, and very volatile, paths
for the nominal money supply (under ‡exible rates). Or viceversa, we would
need to observe some policymakers setting simple paths for the money supply
and others setting the equivalent (and possibly highly volatile) paths for the
nominal exchange rate. This is not how policymakers operate in practice. In
the actual world, policymakers operating under predetermined exchange rates
will set some simple paths for the nominal exchange rates and policymakers
operating under ‡exible exchange rates will set simple paths for the nominal
money supply.27 After all, nominal exchange rate rules or money supply rules
are credible and provide an anchor to in‡ationary expectations provided that
2 6 If, under predetermined exchange rates, the Central Bank were increasing domestic credit

so as to satisfy the increasing money demand, then the two systems would behave identically
in all dimensions.
2 7 We will deal with nominal interest rate rules in Chapter 9 but the same argument would

apply. A given nominal interest rate rule would correspond to very volatile paths of the
nominal money supply and nominal exchange rate, depending on what shocks are hitting the
economy.

21
they are simple and easily understandable by the public. As result, exchange
rate regimes are not equivalent in practice. The next section provides an il-
lustration of this by studying the economy’s response to the same shock under
predetermined and then ‡exible exchange rates. We will see that, as we would
expect, the response is quite di¤erent.

5 Temporary money shocks

This section illustrates the practical “non-equivalence”of exchange rate regimes
by analyzing how an anticipated negative shock to money demand will lead to
quite a di¤erent dynamic response under predetermined and ‡exible exchange
rates. Under predetermined, the in‡ation rate (i.e., the rate of devaluation) is
controlled by policymakers and the negative demand shock will show up as a
loss of reserves. In sharp contrast, under ‡exible rates, in‡ation (i.e., the rate of
devaluation) will increase in anticipation of the shock and reach its maximum
just before the shock actually happens (which is reminiscent of Sargent and
Wallace’s (1981) unpleasant monetarist arithmetic).
To tackle money demand shocks in our framework, we modify preferences in
the following way:
Z 1
[u(ct ) + t v(mt )]e t dt, (44)
0

where t (> 0) is a shock to the liquidity services provided by real money bal-
ances. The rest of the model remains unchanged.
While the …rst-order condition for consumption remains given by (12), the
…rst-order condition for real money balances now reads as:
0
tv (mt ) = it . (45)
Combining (12) and (45) implicitly de…nes a real money demand of the form:

mt = L(ct ; it ; t ); (46)
@L it u00 (c)
= > 0;
@ct v 00 (m) t
@L u0 (c)
= 00
< 0;
@it t v (m)
@L v 0 (m)
= 00
> 0.
@ t t v (m)

As we should have expected, an increase in t raises real money demand:

We will now consider a perfect foresight equilibrium path for a non-constant
path of the money shock parameter, t . (We assume that foreign in‡ation is
constant over time and equal to .) Speci…cally, suppose that t is constant

22
until T at which point it falls. In other words, there is an anticipated negative
money demand shock. Formally:
H
0 t < T;
t = L (47)
t T;

where H > L .
We will now characterize the corresponding PFEP under both predetermined
exchange rates and ‡exible exchange rates.

5.1 Predetermined exchange rates

Suppose that the economy is operating under a predetermined exchange rate
regime with a constant rate of devaluation given by ". The path of the money
shock parameter is given by (47) (Figure 7, Panel A). How will the economy
behave under predetermined exchange rates?

[Figure 7]
We know, of course, that consumption is not a¤ected (Figure 7, Panel B).
The path of the nominal interest rate which, given the interest parity condition
(23), is determined by the constant rate of devaluation, is also constant over
time (Figure 7, Panel F). Real money balances, however, will fall at time T in
response to the negative money demand shock, as follows from the real money
demand (46) (Figure 7, Panel C). Since the real stock of domestic credit remains
constant throughout, we know that the nominal money supply falls at T . This,
in turn, corresponds to a fall in international reserves at T (Figure 7, Panel D).

5.2 Flexible exchange rates

Suppose now that the economy is operating under a ‡exible exchange rate regime
with a constant rate of money growth given by . Again, the path of the money
shock parameter is given by (47) (Figure 8, Panel A). How will this economy
behave over time?

[Figure 8]
Under ‡exible exchange rates, the path of real money demand will be gov-
erned by an unstable di¤erential equation along the lines of (41). Given that we
now have a money demand shock in the model, it is straightforward to verify
that the corresponding di¤erential equation for real money balances is given by
0
tv (mt )
m
_ t = mt r + :

The stationary value of real money demand is therefore implicitly given by:
0
tv (mt )
r+ = . (48)

23
It follows from (48) that at time T , the stationary value of real money balances
falls (recall that is constant along a perfect foresight path). Further, for the
same reasons argued above (i.e., both M and E are continuous functions of
time at T ), m will be continuous at time T . For the path of m to be continuous
at time T , m needs to start below the stationary equilibrium corresponding to
H
, fall over time, and reach the stationary equilibrium corresponding to L at
precisely time T . The path of real money balances is illustrated in Figure 8,
Panel C.
Given the path of real money balances depicted in Figure 8, Panel C, we
can derive the path of the rate of depreciation and hence of the nominal interest
rate. Since m _ t =mt = "t , it follows that
m
_t
"t = . (49)
mt
Equation (49) tells us four critical things. First, for t T; m _ t = 0 and therefore
"t will be constant and equal to . Second, for t 2 [0; T ), the fact that
m_ t < 0 implies that "t > . Third, at time T "t will jump down because,
as Figure 8, Panel C makes clear, m _ t is negative until just before T and then
becomes zero at T . Fourth, to …nd out how "t behaves for t 2 [0; T ), di¤erentiate
(49) with respect to time to obtain:
1
"_ t = (m
• t mt _ 2t ) > 0;
m
m2t
where the sign follows from the fact that m • < 0. Putting together all these
pieces of information, we conclude that "t starts above its stationary equilibrium,
increases over time, and then falls at time T (Figure 8, Panel D). Given interest
parity, the path of the nominal interest rate will follow exactly the same pattern
(Figure 8, Panel F).
What about the nominal exchange rate? We already know that the nominal
exchange rate cannot jump at T . Hence, given that the rate of depreciation
itself increases over time, the logarithm of the nominal exchange rate increases
over time at an increasing rate and then becomes constant at time T (Figure 8,
Panel E.)
The remarkable feature of this case is that the in‡ation rate is high (and
increasing over time) before the negative money demand shock actually occurs.
In fact, an outside observer who is not aware of the forthcoming shock to money
demand would be quite dumbfounded by the emergence of in‡ation when the
money supply has not changed at all!
In sum – and comparing Figures 7 and 8 – we can see the very di¤erent
response of this economy to an anticipated fall in money demand. Under pre-
determined exchange rates, this shock has no e¤ect whatsoever on the in‡ation
rate (i.e., on the rate of devaluation) or the nominal interest rate. In sharp
contrast, under ‡exible exchange rates, the rate of in‡ation (i.e., depreciation)

24
rises in anticipation of the shock, as does the nominal interest rate.28 ’29

6 Money as a veil in a cash-in-advance model

So far in this chapter we have examined monetary phenomena in a money-in-
the-utility function model in which money is a veil in the sense that the real
equilibrium is independent of monetary/exchange rate policy. Another popular
way of introducing money into the endowment model of Chapter 1 is through a
cash-in-advance constraint.30 We will now study a discrete-time formulation of
a cash-in-advance model in which money is a veil and analyze how the monetary
equilibrium is determined under both predetermined and ‡exible exchange rates.
In so doing, we will also have a chance to go over some important accounting
in discrete-time terms.

6.1 Households
6.1.1 Budget constraints
We …rst need to be speci…c about the economic environment in which households
operate. Households enter period t with a certain amount of nominal cash
balances, Mt 1 , and a certain amount of nominal foreign bonds, Bt 1 . Periods
are divided into two sub-periods. In the …rst sub-period, asset markets open.
In the asset markets, agents receive/pay interest on the net foreign bonds they
carried over from last period, buy or sell bonds in exchange for money (at the
Central Bank under predetermined exchange rates or in the foreign exchange
market under ‡exible exchange rates) and receive nominal transfers from the
government, Pt t . Households exit the asset market with a quantity Mtp of
nominal cash balances and Bt of nominal bonds. Formally:

Mtp + Et Bt = Et (1 + it 1 )Bt 1 + Mt 1 + Pt t . (50)

In the second sub-period, goods market open. Think of households as com-
posed of two individuals: a shopper and a seller. The shopper and seller part
at the beginning of the goods market sub-period and do not meet again until
goods markets close.31 Think of the seller as staying in the store selling the
endowment of the good to other households’shoppers. The shopper leaves the
store with nominal money balances of Mtp and uses part or all of this money to
buy goods from other stores. Since, by assumption, the shopper needs to use
2 8 Drazen and Helpman (1990) study the in‡ationary consequences of anticipated policies

in the context of a closed economy.

2 9 Exercise 5 at the end of this chapter provides yet another illustration of how the economy

responds di¤erently depending on the exchange rate regime by studying the consequences of
an anticipated increase in the rate of growth of the nominal anchor.
3 0 See Helpman (1981) and Lucas (1982).
3 1 If it helps you, think of a “period” as a day. Assets markets open in the morning and

close at noon. Goods markets open at noon and close at 5 pm. The shoppper and the seller
say good-bye at noon and do not see each other until after 5 pm.

25
money acquired in the asset markets to buy goods in the goods markets, we
refer to this as a cash-in-advance constraint.32 Formally:

Mtp Pt ct . (51)
What are the households’s money balances at the end of period t (denoted
by Mt )? Households will have the money obtained from selling goods at the
store (Pt yt ) and the money brought from the asset markets that was not spent
on purchasing goods (Mtp Pt ct ). Formally,

Mt = Mtp Pt ct + Pt yt . (52)
By substituting (52) into (50), we obtain the households’ ‡ow constraint for
period t as a whole:

Mt + Et Bt = Et (1 + it 1 )Bt 1 + Mt 1 + Pt t + Pt yt Pt ct . (53)
For the sake of comparison with the continuous-time case, we can de…ne nominal
assets as At Mt + Et Bt and, by adding and subtracting Et 1 Bt 1 in the LHS
of equation (53), rewrite it as:

At At 1 = Et it 1 Bt 1 + (Et Et 1 )Bt 1 + Pt t + Pt yt Pt ct ,
| {z } | {z }
interest paym ents capital gains

which is the discrete-time counterpart of ‡ow constraint (5) in the continuous-

time case.
To express the ‡ow constraint in real terms, divide both sides of (53) by Pt
and manipulate terms to obtain:
Et Pt 1
mt + bt = P (1 + it 1 )bt 1 + mt 1 + t + yt ct ;
Pt t 1 Pt
where, by de…nition, mt Mt =Pt and bt Et Bt =Pt = Bt =Pt denote real
money balances and real bond holdings. De…ning the in‡ation rate in period t
as 1 + t = Pt =Pt 1 , assuming – as usual – that the Fisher equation holds in
the rest of the world (i.e., 1 + it 1 = (1 + r)(Pt =Pt 1 )), and using the law of
one price (i.e., Pt = EPt ), we can rewrite the last equation as
mt 1
bt + mt = (1 + r)bt 1+ t + yt ct :
+ (54)
1+ t
Adding and subtracting (1 + r)mt 1 from the RHS of the last equation and
taking into account that, by de…nition, at = mt + bt , we obtain:
it 1
at = (1 + r)at 1 + t + yt ct mt 1, (55)
1+ t
3 2 The implicit assumption is that even though all households are identical, they do not

consume their own endowment. If it helps, think of each households as being endowed with
the same good (say, candy) but that candy comes in di¤erent colors. Households do not like
the color of their own candy and wish to buy other household’s candies.

26
which is the discrete-time counterpart of equation (9). To understand intuitively
the opportunity cost term (i.e., the last term on the RHS), notice that by
holding nominal money balances given by Mt 1 from period t 1 to period t,
the household foregoes interest payments in the amount of it 1 Mt 1 , the real
value of which is it 1 Mt 1 =Pt in period t. By multiplying and dividing by Pt 1 ,
this term can be expressed as it 1 mt 1 =(1 + t ).

6.1.2 Utility maximization

We can set up the household’s maximization problem in various ways depending
on which constraints we use. It proves convenient to substitute equation (50)
into the cash-in-advance constraint (51) to obtain:

Et (1 + it 1 )Bt 1 + Mt 1 + Pt t Et B t Pt ct . (56)
Expressing this equation in real terms, we obtain:
mt 1
(1 + r)bt + t bt ct .
1 + (57)
1+ t
Households then choose fct ; bt; mt g1
t=0 to maximize lifetime utility subject to
a sequence of ‡ow budget constraints given by (54) and a sequence of inequal-
ity constraints given by the cash-in-advance constraints (57). In terms of the
Lagrangian,

1
X
t
L= u(ct )
t=0
X1
t mt 1
+ t (1 + r)bt 1 + + t + yt ct bt mt
t=0
1+ t

X1
t mt 1
+ t (1 + r)bt 1 + + t bt ct .
t=0
1+ t

The …rst-order conditions for ct , mt , and bt are given, respectively, by

u0 (ct ) = t + t; (58)
t+1 t+1
t + + = 0; (59)
1 + t+1 1+ t+1
(1 + r)( t+1 + t+1 ) = t + t: (60)
The …rst-order condition for t recovers equation (54). The …rst-order con-
dition for t takes the form of a Kuhn-Tucker condition:

mt 1 mt 1
(1+r)bt 1+ + t bt ct t 0; (1 + r)bt 1 + + t bt ct t = 0:
1+ t 1+ t
(61)

27
In words, if the cash-in-advance holds as an inequality, the associated multiplier
will be zero.
Given the assumption that (1 + r) = 1, it follows from condition (60) that
t+1 + t+1 = t + t and hence, from condition (58), that

u0 (ct+1 ) = u0 (ct ): (62)

As in the endowment economy of Chapter 1, consumption will be constant
along any perfect foresight path. Moreover, since we have yet to say anything
about monetary/exchange rate policy, full consumption smoothing will hold
irrespective of the exchange rate regime in place. This proves that, in this
version of the cash-in-advance model, money is a veil in the sense that the
real equilibrium will not depend on the path of monetary variables. Hence,
in respond to ‡uctuations in the endowment, the economy will use the trade
balance as a shock absorber and keep consumption constant by borrowing in
bad times (i.e., when the endowment is low) and repaying in good times (i.e.,
when the endowment is high).
We will now show that if the nominal interest rate is positive, the cash-in-
advance constraint will bind. Taking into account that = 1=(1 + r), condition
(59) becomes

t+1 + t+1
= t:
1 + it
Since t+1 + t+1 = t + t, we can rewrite this last condition as:

t it = t:

Since t > 0, it follows that if it > 0, then t > 0 which implies, from the
slackness condition, that the cash-in-advance binds. This is, of course, very
intuitive. If the nominal interest rate is positive, it makes no sense for households
to leave the asset markets with more money than they need to purchase goods
when goods markets open because they could have always used that money to
buy bonds and receive interest rate payments at the beginning of the following
period.
On the other hand, if it = 0, t = 0 and the cash-in-advance constraint is
non-binding in the sense that it becomes irrelevant for the consumer’s choice.
In this case, households will be indi¤erent between holding money or bonds and
hence the choice of money balances is indeterminate.33

6.2 Government
As before, let H denote the foreign currency value of international reserves.
The government’s ‡ow budget constraint in nominal terms is thus given by
3 3 Of course, from a purely mathematical point of view, the cash-in-advance constraint could

still hold as an equality even if t = 0. But, from an economic point of view, this is irrelevant
since money and bonds are perfect substitutes and the cash-in-advance does not constrain
consumers’choices.

28
 Et Ht = Et (1 + it 1 )Ht 1 + Mt Mt 1 Pt t .
To express it in real terms, divide both sides of this equation by Pt to obtain
Mt Mt 1
ht = (1 + r)ht 1 + t, (63)
Pt
where we have used the fact that, by de…nition, ht Ht =Pt , from the law of
one price, Pt = Et Pt , and, from the Fisher equation in the rest of the world,
1 + it 1 = (1 + r)(Pt =Pt 1 ).
Finally, and for further reference, notice that revenues from money creation
may be expressed as follows:
Mt Mt 1 t
= mt mt 1 + mt 1: (64)
Pt 1+ t

6.3 Equilibrium conditions

Perfect capital mobility implies that interest parity holds:
Et+1
1 + it = (1 + it ) . (65)
Et
To obtain the economy’s ‡ow constraint, combine the consumer’s ‡ow constraint
(given by equation (54)) with the government’s (given by equation (63)) –taking
into account (64) –to obtain:

bt + ht = (1 + r)(bt 1 + ht 1) + yt ct . (66)
Let k( b + h) denote the economy’s total net foreign assets. Iterating
kt
forward and imposing the transversality condition lim (1+r) t = 0 yields:
t!0
1
X 1
X
ct yt
t = (1 + r)(b 1 +h 1) + t: (67)
t=0 (1 + r) t=0 (1 + r)

6.3.1 Perfect foresight equilibrium

We have established from (62) that consumption is constant over time. From
the resource constraint (67), it then follows that:
" 1
#
r X yt
c= (1 + r)(b 1 + h 1 ) + t : (68)
1+r t=0 (1 + r)

As in the continuous-time case, the constant level of consumption is equal to

permanent income.
Taking into account (51) and the fact that we have established that the CIA
constraint will bind, the constant value of consumption will determine the value
of real money balances taken into the goods markets:

29
 Mtp
= c: (69)
Pt
Further, notice that substituting (69) into (52), it follows that

Mt = Pt yt , (70)
which you will recognize as a quantity theory equation with unitary velocity.

6.4 Predetermined exchange rates

Under predetermined exchange rates, the Central Bank sets the path of the
nominal exchange rate (i.e., E0 , E1 , ...) and the path of the stock of nominal
domestic credit (D0 , D1 , ...). Speci…cally –and as in the continuous-time case
– suppose that the monetary authority sets E0 and then a constant rate of
devaluation. Formally,
Et+1
= 1 + "; t = 0; 1; 2; ::::
Et
Given this constant rate of devaluation, the nominal interest rate follows from
the interest parity condition (65):

1 + it = (1 + i )(1 + "). (71)

Let us now determine the path of the price level. Let the rate of foreign
in‡ation be constant; that is: Pt+1 =Pt = 1 + . Then, given the law of one
price (P = EP ), the initial price level is given by P0 = E0 P0 . The law of one
price also determines a constant level of in‡ation:

1+ = (1 + ") (1 + );
where 1 + = Pt+1 =Pt .
We now turn to the path of the nominal money supply. Given the quantity
theory equation (70), M0 = P0 y0 . The rate of growth of the nominal money
supply also follows from the quantity theory equation (70):
yt+1
1+ t+1 = (1 + ) ;
yt
where 1 + t+1 Mt+1 =Mt . Money supply growth will be higher in good times
and lower in bad times because a higher (lower) endowment implies higher
(lower) sales during the goods market period and hence more (less) nominal
money balances carried over to next period.
To derive the path of international reserves, start from the Central Bank’s
balance sheet, given by
Dt Mt
ht + = .
Et Pt Et Pt

30
Considering the same identity for t + 1 and subtracting, we get:

Mt+1 Mt Dt (1 + ) (1 + ")(1 + )
ht+1 ht = ;
Et+1 Pt+1 Et Pt Et Pt (1 + ")(1 + )

where is the constant rate of domestic credit set by the Central Bank. For
the same reasons discussed for the continuous-time case, we will assume that
1 + = (1 + ")(1 + ) to ensure that the predetermined exchange rate system
is sustainable over time. Imposing that assumption, we can rewrite this last
expression as
Mt+1 Mt
ht+1 ht = ;
Et+1 Pt+1 E t Pt
which is, of course, the counterpart to equation (38) in the continuous-time
case. Changes in international reserves will solely re‡ect changes in real money
balances. Using the quantity theory equation (70), we can rewrite this last
equation as

ht+1 ht = yt+1 yt :
Increases (decreases) in the endowment will be re‡ected in higher (lower) re-
serves through their e¤ect on real money demand. This is a feature that is not
present in the continuous-time case because in that case the relevant variable
for price level determination is consumption and not output.
This characterization of predetermined exchange rates makes clear the di-
chotomy between the real and monetary sectors of the economy. Speci…cally –
and since we made no assumption on exchange/monetary policy in deriving the
constant level of consumption (68) – exchange rate policy is both neutral and
superneutral.

6.5 Flexible exchange rates

Under ‡exible exchange rates, the Central Bank controls the path of the money
supply by setting Mt , t 0. Speci…cally, we assume that the monetary authority
sets M0 and then sets a constant rate of money growth:

Mt = (1 + )Mt 1, t 1:
The rate of in‡ation follows from the quantity theory equation, given by expres-
sion (70):
Pt+1 1+
= ; t 0. (72)
Pt yt+1 =yt
Since the quantity theory equation holds, of course, for t = 0, the initial price
level is determined by

31
 M0
P0 = .
y0
The nominal exchange rate is determined by the law of one price:
Pt
Et = : (73)
Pt
The rate of depreciation is thus (once again, assume a constant foreign in‡ation
rate)

Et+1 Pt+1 =Pt

= : (74)
Et 1+
From (65), (72), and (74), the nominal interest rate is thus given by

1+i 1+
1 + it = :
1+ yt+1 =yt
Under ‡exible exchange rates, ‡uctuations in the endowment will be re‡ected in
‡uctuations in in‡ation, the rate of depreciation, and the nominal interest rate.
Under predetermined exchange rates, the same ‡uctuations would be re‡ected
in movements in international reserves.
Once again, it is easy to see that monetary policy would be both neutral
and superneutral.

6.6 Concluding remarks

As discussed before, in this basic monetary model, money is a “veil” in the
sense that changes in monetary/exchange rate policy do not a¤ect the path of
real variables. While this model provides the natural conceptual benchmark for
studying monetary economics in the open economy –and could even be a good
description of the actual world in the long run or under extreme hyperin‡ation-
ary conditions – it certainly does not provide us with tools to understand the
possible real e¤ects of monetary/exchange rate policy in an open economy. In
essence, the main task of monetary economics in the open economy is to study
departures from this benchmark in which monetary and exchange rate policy
have real e¤ects.
Subsequent chapters will thus introduce various frictions that will remove the
veil and allow monetary/exchange rate policy to a¤ect the real sector. Specif-
ically, Chapter 6 will assume that there are no interest-bearing bonds in the
economy; Chapter 7 will introduce money through a continuous-time cash-in-
advance constraint that will establish a link between nominal interest rates and
consumption; and Chapter 8 will introduce sticky prices.

32
7 Appendices
7.1 Breaking down the government into the monetary and
the …scal authority
To simplify the presentation of the basic monetary model, we considered in the
text the government as a whole and did not break it down into its separate
entities (i.e., the monetary and the …scal authority). It proves illuminating,
however, to consider each entity separately and see how aggregating them leads
to equation (19) in the text.

7.1.1 The monetary authority

The Central Bank holds international reserves, prints money, lends to the gov-
ernment by issuing domestic credit (think of domestic credit as loans to the
government), and gives transfers to the …scal authority. As in the text, let Ht
denote the net foreign assets (measured in the foreign currency) held by the
monetary authority and Ht denote the domestic currency value of these inter-
national reserves (i.e., Ht Et Ht ). The Central Bank’s ‡ow budget constraint
in domestic currency terms is then given by
g
H_ t = it Et Ht + E_ t Ht + it Dt + M_ t (D_ + Pt t ). (75)
| {z } | {z }
revenues exp enditures

The …rst four terms on the RHS of equation (17) represent sources of rev-
enues for the Central Bank. Speci…cally, the …rst term (it Et Ht ) captures the
domestic-currency value of the interest proceeds on the stock of international
reserves; the second term (E_ t Ht ) denotes the capital gains on the existing stock
of international reserves; the third term (it Dt ) captures the interest income on
the stock of nominal domestic credit; and the fourth term (M_ t ) indicates that
money printing is a source of revenues for the Central Bank. The last two
terms on the RHS of equation (17) capture the Central Bank’s expenditures.
The Central Bank buys domestic bonds issued by the …scal authority (D_ t ) and
transfers to the …scal authority whatever pro…ts it makes (Pt gt ).34
Dividing (75) by Pt –and recalling that Pt = Et Pt –we obtain
:
H_ t M_ t D_ t g
= it ht + "t ht + it dt + t; (76)
Pt Pt Pt
where ht Ht =Pt .
Recall that

H_ t
h_ t = ("t + t )ht . (77)
Pt
3 4 As will become clear below, the standard convention is to assume that the Central Bank

does not accumulate/decumulate wealth (i.e., it keeps its net worth constant over time).

33
Substituting (76) into (77) (and imposing the Fisher equation for the rest of the
world), we obtain:

M_ t D_ t g
h_ t = rht + it dt + t. (78)
Pt Pt
We can rewrite this expression as:
g
h_ t + d_t m
_ t = rht + it dt + ("t + t )mt ("t + t )dt t. (79)
By convention, we assume that the Central Bank’s net worth (i.e., its capital)
remains constant over time. In other words, h_ t + d_t m _ t = 0. Imposing this
condition in the last expression and rearranging terms, we obtain
g
t = it mt . (80)

Intuitively, the Central Bank transfers to the …scal authority its “pro…ts”so as to
keep its net worth constant. The Central Bank’s pro…ts derive from the fact that
while its assets bear market interest rates (remember that both the international
reserves and the stock of domestic credit bear interest), its liabilities (the money
supply) bear no interest. Alternatively, using the Central Bank’s balance sheet
(mt = ht + dt ) and the interest parity condition (it = r + t + "t ), the last
expression can be rewritten as
g
t = rht + rdt + ("t + t )mt .

In this interpretation, the Central Bank’s pro…ts consist of the real interest rate
on reserves and domestic credit and the in‡ation tax on the real money supply.

7.1.2 The …scal authority

The …scal authority (in practice the Finance Ministry) borrows from the Central
Bank (i.e., sells bonds to the Central Bank), pays interest on these bonds (i.e., on
the stock of domestic credit), receives a transfer of gt from the Central Bank,
and makes lump-sum transfers to the private sector.35 The …scal authority’s
‡ow constraint in nominal terms is thus

D_ t = it Dt + Pt t Pt gt : (81)
| {z } | {z }
exp enditures revenues

Dividing this expression by Pt , we get

D_ t g
= it dt + t t: (82)
Pt
Using (80) and rearranging terms, we obtain
3 5 Needless to say, the …scal authority would typically spend on goods (as in chapter 4),

a feature that we are abstracting from in this chapter and to which we will return in later
chapters.

34
 t = d_t + ("t + t )mt + rht .
We see that if d_t = 0 (the usual assumption), then t = ("t + t )mt + rht ,
which coincides of course with what we derived in the text (equation (40) in the
case of predetermined exchange rates and equation (43) in the case of ‡exible
rates, in which case ht = 0 for all t)).

7.1.3 Aggregating the monetary and the …scal authority

Combining the monetary’s authority nominal budget constraint –given by equa-
tion (75) –and the …scal authority’s nominal budget constraint –given by equa-
tion (5) –we obtain

H_ t = it Et Ht + E_ t Ht + M_ t Pt t , (83)
which, of course, coincides with equation (17) in the text.
By the same token, by combining the monetary authority’s ‡ow constraint
–given by (78) –and the …scal authority’s ‡ow constraint –given by (82) –we
obtain:

M_ t
h_ t = rht + t, (84)
Pt
which coincides with constraint (19) in the text. As a …nal remark, notice that
we have just gone over a purely accounting exercise so that all the derivations
in this appendix hold for any exchange rate regime.

7.2 Accounting in basic monetary model with jumps in

real money balances
We mentioned in the text that, strictly speaking, both the household’s and
the government’s budget constraints should take into account the possibility of
jumps in real money balances (see, for example, Drazen and Helpman (1987)).
We develop such a formulation in this appendix.

7.2.1 Consumer
The consumer’s ‡ow constraint takes the form
bt bt = (Mt Mt )=Et ; if t 2 J;
(85)
b_ t = rbt + yt + t ct m
_ t "t mt ; if t 2
= J:

These ‡ow constraints allow for the possibility of discrete changes in bt and mt
at a …nite set of points (which may include t = 0) belonging to the set J.36
The precise points that belong to set J will naturally depend on the speci…c
3 6 A jump at t = 0 may occur in the case of an unanticipated shock at t t = 0. If we are

analyzing a PFEP, then, by construction, there will be no jumps at t = 0.

35
problem at hand. For example, when we analyze the e¤ects of an unanticipated
and permanent shock under predetermined exchange rate, the set J will typically
include two points: t = 0 and t = T . Notice also that, along a perfect foresight
path, the total level of …nancial assets, at , cannot change discretely at any point
in time.
Integrating forward and imposing the condition lim e rt bt = 0, we obtain
t!1
the following consumer’s lifetime budget constraint:
Z 1 Z 1 X Mt Mt
b0 + (yt + t ) e rt dt = (ct + m_ t + "t mt )e rt dt + e rTj ;
0 0 Et
j2J
(86)
where b0 denote net foreign assets an instant before t = 0 and Tj are values of
t that belong to set J.
This intertemporal constraint can be further simpli…ed if we impose the
condition that lim e rt mt = 0 and use the interest parity condition (it = r +"t )
t!1
to obtain:37
Z 1 Z 1
M
b0 + 0 + (yt + t) e
rt
dt = (ct + it mt )e rt
dt: (87)
E0 0 0
Unless there is a discrete change in the consumer’s real …nancial assets at t = 0,
then b0 + M0 =E0 = a0 and this expression coincides with constraint (10) in
the text. Even if there is a discrete change in the consumer’s real …nancial
assets at t = 0 – which would be the case if an unanticipated devaluation
occur –our assumption that the government transfers back to the consumer all
proceeds ensures that the solution would be the same as in the text. Of course,
if the government used the proceeds of a discrete devaluation to, say, increase
spending, we would need to modify the analysis.

7.2.2 Government
The government’s ‡ow constraint should read as

ht ht = (Mt Mt )=Et ; if t 2 J;
(88)
h_ t = rht + m
_ t + "t mt t. if t 2
= J;
where the set J has been de…ned above. Under predetermined exchange rates,
these discrete changes will take place when consumers decide to trade net foreign
assets for domestic money (or viceversa) at the Central Bank:
If we integrate forward equation (88), imposing the condition that lim e rt ht =
t!1
0, we obtain the following intertemporal constraint for the government:

Z 1 Z 1 X Mt
rt : rt Mt rTj
te dt = h0 + (mt + "t mt )e dt + e , (89)
0 0 Et
j2J

3 7 This condition will hold in equilibrium because, at an optimum, the choice of m will be
t
…nite.

36
where h0 denotes the level of international reserves an instant before t =
0. Notice that a, say, discrete fall in real money balances implies a loss of
revenues for the government. This intertemporal constraint simply says the
present discounted value of transfers must be …nanced with the initial stock of
international reserves plus the present discounted value of revenue from money
creation (including discrete jumps).
This intertemporal constraint can be further simpli…ed if we impose the
condition that lim e rt mt = 0 and use the interest parity condition (it = r +"t )
t!1
to obtain:38
Z 1 Z 1
rt M0 rt
te dt = h0 + it mt e dt, (90)
0 E0 0

7.2.3 Aggregation
Combining the consumer’s intertemporal constraint, given by equation (87),
with the government’s, given by equation (90), we obtain
Z 1 Z 1
b 0 + h0 + yt e rt dt = ct e rt dt:
0 0

Unless there is a jump in the economy’s net foreign assets at t = 0, then b0 +

h0 = k0 and this intertemporal constraint coincides with expression (26) in
the text.

7.3 Restrictions on the rate of growth of domestic credit

In deriving the government’s intertemporal constraint –given by equation (22)
–we imposed the condition
rt
lim ht e = 0: (91)
t!1

Similarly, in deriving the economy’s resource constraint –given by equation (26)

–we imposed the condition
rt
lim kt e = 0: (92)
t!1

Since kt = ht + bt , conditions (91) and (92) imply that

rt
lim bt e = 0: (93)
t!1

Further, in deriving the consumer’s intertemporal constraint –given by equation

(10) –we imposed the condition
rt
lim (bt + mt )e = 0: (94)
t!1
3 8 This condition will hold in equilibrium because, at an optimum, the choice of m will be
t
…nite.

37
Conditions (93) and (95) imply that
rt
lim mt e = 0: (95)
t!1

Since, from the Central Bank’s balance sheet, dt = mt ht , conditions (91)

and (95) imply
rt
lim dt e = 0: (96)
t!1

By de…nition dt = Dt =Et Pt . Hence,

Rt
( "s s )ds
dt = d0 e 0 s
:

Using the latter equation, we can rewrite (96) as

Rt
( "s r)ds
lim d0 e 0 s s = 0: (97)
t!1

This condition provides an upper bound on the rate of expansion of domestic

credit which is consistent with the maintenance of a predetermined exchange
rate regime. In the case in which s , "s , and s are constant over time and
equal to , ", and , respectively, then the condition = " + , as assumed in
the text, is su¢ cient to ensure that (97) holds. In general, however, all that we
need for condition (97) to hold is that, “on average”, s "s s r < 0. In the
case in which the exchange rate is …xed ("t = 0), foreign in‡ation is zero, and
the rate of growth of domestic credit is constant at , then condition (97) holds
as long as < r.39 . Intuitively, under a …xed exchange rate, as long as > 0, the
Central Bank is increasing its external debt (i.e., international reserves become
increasingly negative) but can …nance the debt service by lump-sum taxing the
private sector. If, however, > r, it is no longer possible to service the debt
because the rate of growth of government’s debt exceeds the real interest rate.
As Obsfeld (1986) points out, if lump-sum taxation were not available, then any
> 0 is no longer feasible because it would imply ever-growing revenue needs
that can only be raised by distortionary taxation.

7.4 MIUF model in discrete time

This appendix develops a discrete-time version of the continuous-time MIUF
model developed in the text. This will prove useful for subsequent chapters in
the book. Unlike real models in which the switch from continuous to discrete
time does not pose any particular methodological problem, in monetary models
this switch is non-trivial because – as will become clear below – timing issues
play a critical role.
3 9 Thisis one of Obstfeld’s (1984) points: if there are no constraints on the government’s
external borrowing, then a …xed exchange rate regime is sustainable as long as the rate of
growth of domestic credit does not exceed the world real interest rate.

38
7.4.1 Households’maximization
Suppose that preferences are given by
1
X
t Mt
u(ct ) + v : (98)
t=0
Pt
A critical issue in discrete-time monetary models is related to timing issues. The
more common assumption in the literature for MIUF models –which is captured
in the above preferences –is that end-of-period money balances enter the utility
function. While this assumption may look somewhat odd, it is the most common
assumption in the literature (see, for example, Calvo and Leiderman (1992)).40
The intertemporal budget constraint remains given by (55).
Households thus choose fct ; mt ; at g1
t=0 to maximize lifetime utility – given
by (98) – subject to a sequence of ‡ow budget constraints given by (55). In
terms of the Lagrangian:

1
X
t
L= [u(ct ) + v (mt )]
t=0
X1
t it 1
+ t (1 + r)at 1 + t + yt ct mt 1 at .
t=0
1+ t

The …rst-order conditions with respect to ct ; mt ; and at are given by, re-
spectively,

u0 (ct ) t = 0; (99)
it
v 0 (mt ) t+1 = 0; (100)
1 + t+1
t + (1 + r) t+1 = 0; (101)

Notice that, under our usual assumption that (1+r) = 1, the last condition
reduces to:

t+1 = t.

The Lagrange multiplier is constant over time. Denote the constant value of the
multiplier by (i.e., t+1 = t = ). Hence, from (99), it follows that

u0 (ct ) = : (102)
As expected, consumption is constant over time.
4 0 As discussed by Calstrom and Fuerst (2001), the assumption is rather odd because it

amounts to assuming that what matters for liquidity purposes is the money balances that
the consumer leaves the grocery store with, rather than the money that he/she enters the
store with! The more natural assumption would be that beginning-of-period money balances
provide liquidity services, a speci…cation analyzed below.

39
 Taking into account that (i) the Lagrange multiplier is constant; (ii) (1 +
r) = 1 and (iii) by de…nition, 1+it = (1+r)(1+ t+1 ), we can rewrite …rst-order
condition (100) as
it
v 0 (mt ) = : (103)
1 + it
This is the condition analogous to equation (13) in the continuous-time version.
As equation (103) makes clear, the opportunity cost of holding real money bal-
ances is it =(1 + it ). Intuitively, in period t the household uses Mt to acquire
money balances (instead of purchasing interest-bearing bonds) and hence fore-
goes it Mt in interest payments at the beginning of period t + 1. The real value
of these interest payments in t + 1 is it Mt =Pt+1 , which discounted to time t
amounts to it (Mt =Pt+1 )=(1 + r). Multiplying and dividing by Pt and recalling
that, by de…nition, 1 + it = (1 + r)(Pt+1 =Pt ), this expression can be written as
[it =(1 + it )]mt .
Substituting (102) into (103), we obtain
it
v 0 (mt ) = u0 (ct ) ; (104)
1 + it
which implicitly de…nes a real money demand with standard properties.

7.4.2 Perfect foresight equilibrium

The aggregate conditions (66) and (67) still hold in this model. Equation (102)
tells us that consumption is constant over time. Hence, from the resource con-
straint,
" 1
#
r X yt
c= (1 + r)(b 1 + h 1 ) + t : (105)
1+r t=0 (1 + r)

7.4.3 Predetermined exchange rates

We now solve the model under predetermined exchange rates. Needless to say,
the same logic that applied to the continuous-time case remains valid. The mon-
etary authority sets the path of both the nominal exchange rate and the stock
of domestic credit. Formally, the monetary authority sets Et , t = 0; 1; :::and
Dt ; t = 0; 1; :::Suppose, for simplicity, that the monetary authority sets a con-
stant rate of devaluation. Formally,
Et+1
= 1 + ".
Et
Through the interest parity condition, the constant rate of devaluation deter-
mines a constant nominal interest rate (assuming, of course, a constant foreign
in‡ation rate):
_
1 + i = (1 + i )(1 + "):

40
Given constant consumption and a constant nominal interest rate, real money
demand will also be constant. Given a constant real money demand, the path
of international reserves will be given by

Dt (1 + ) (1 + ")(1 + )
ht+1 ht =
Et Pt (1 + ")(1 + )
Imposing the condition that the predetermined exchange rate regime be sus-
tainable over time (i.e., 1 + = (1 + ")(1 + )), we obtain

ht+1 = ht .
International reserves will be constant over time.

7.4.4 Flexible exchange rates

We now solve the model under ‡exible exchange rates. Assume, for simplicity,
that the monetary authority sets an initial level of the nominal money supply,
M0 , and a constant rate of money growth:
Mt+1
=1+ ; t = 0; 1; ::: (106)
Mt
We will proceed in an analogous way to the continuous-time case and derive
an unstable di¤erence equation for real money balances. To this e¤ect, multiply
and divide by Pt+1 and Pt on the LHS of equation (106) to obtain:
mt+1 1
= Pt+1
(1 + ) : (107)
mt
Pt

Using the interest parity condition and the law of one price, we can rewrite
this equation as

mt+1 (1 + ) (1 + r)
= : (108)
mt 1 + it
Solving for it from (103) and substituting the resulting expression into the last
equation, we obtain a di¤erence equation in mt ,

v 0 (mt )
mt+1 = mt (1 + ) (1 + r) 1 : (109)

This di¤erence equation is the analog to the di¤erential equation (41) in the
continuous-time case.
The stationary state corresponding to this equation is implicitly given by
(as follows from setting mt+1 = mt = m)

(1 + ) (1 + r) 1
v 0 (m) = :
(1 + ) (1 + r)

41
 By linearizing this equation around the stationary state, we can check that
this is an unstable di¤erential equation:41

@mt+1 (1 + ) (1 + r)
=1 m v 00 (m) > 1.
@mt ss

Given the instability of (109), mt will need to be at its stationary state from
t = 0 onwards. Otherwise, it would diverge over time. In other words, the
constant value of mt , t = 0; 1; ::: will be given by

(1 + ) (1 + r) 1
v 0 (m) = u(c) :
(1 + ) (1 + r)
It then follows from (108) that the nominal interest rate will also be constant
over time and given by

it = (1 + ) (1 + r) 1:
Through interest parity, it then follows that the rate of depreciation will also
be constant.

7.5 Beginning-of-period money balances

While it is, by far, the most popular MIUF speci…cation in the literature, the
end-of-period money balances speci…cation is, to say the least, conceptually
problematic. A more natural formulation would have beginning-of-period money
balances provide liquidity services during the current period. In this case, pref-
erences would be given by
1
X
t Mt 1
u(ct ) + v : (110)
t=0
Pt

(Below it will prove convenient to express Mt 1 =Pt as mt 1 (Pt 1 =Pt ).) The
budget constraint remains unchanged. Households thus choose fct ; mt ; at g1 t=0
to maximize lifetime utility – given by (110) – subject to a sequence of ‡ow
budget constraints given by (55). In terms of the Lagrangian:

1
X
t Pt 1
L = u(ct ) + v mt 1
t=0
Pt
X1
t it 1
+ t (1 + r)at 1 + t + yt ct mt 1 at .
t=0
1+ t

The …rst-order conditions with respect to ct ; mt ;and at are given by, respectively
(note that, relative to the end-of-period case, the only …rst-order condition that
4 1 Asin the continous case, if v(m) = log(m), this di¤erence equation becomes a linear
di¤erence equation given by mt+1 = (1 + ) (1 + r) (mt 1= ).

42
changes is the one for real money balances):
u0 (ct ) t = 0; (111)
Pt Pt it
v 0 mt t+1 = 0
Pt+1 Pt+1 1 + t+1
t + (1 + r) t+1 = 0;
As in the previous case, the last equation implies that t+1 = t and hence,
from (111), that consumption will be constant over time. In turn, the …rst-order
condition for real money balances simpli…es to:

Mt 1
v0 = t it 1 , (112)
Pt
which makes clear that the opportunity cost of the relevant measure of real
money balances is simply it 1 . Intuitively, if last period the consumer had
bought bonds –instead of holding money –he/she would have received it 1 Mt 1
at the beginning of period t, the real value of which is it 1 Mt 1 =Pt .
It would be straightforward to solve the model under predetermined ex-
change rates. We solve instead the ‡exible exchange rates case. Once again,
we will proceed by deriving an unstable di¤erence equation. To this e¤ect, it
proves convenient to de…ne a new measure of real money balances, denoted by
nt , as
Mt 1
nt .
Pt
Notice that the variable nt captures the measure of real money balances that
is relevant for period t purposes. In light of this de…nition, we can rewrite
condition (112) as

v 0 (nt ) = t it 1 . (113)
To derive the di¤erence equation, start from the fact that Mt =Mt 1 = 1 + ,
multiply and divide on the LHS by Pt and Pt+1 , and use the interest parity
condition to obtain:

nt+1 (1 + ) (1 + r)
= .
nt 1 + it
Solve for it from (113) (iterated forward one period) and substitute the resulting
expression into the last equation to obtain, after rearranging terms,

nt+1 (1 + ) (1 + r)
= 0 ,
nt 1 + v (nt+1 )
which implicitly de…nes a di¤erence equation in nt . Moreover, since
dnt+1 1
= v 00 (n)
> 1;
dnt 1 + n h v)(1+r)
(1+
i
0 (n) 2
1+

43
this di¤erence equation is unstable. We thus conclude that the path of nt will
be constant over time. Proceeding as before, we can easily characterize the rest
of the system.

44
 Exercises42

1. Dirty ‡oating
This exercise illustrates how one would think about “dirty ‡oating”in the
monetary model analyzed in the main section. Speci…cally, we analyze
how the economy would respond to positive monetary shock that would
lead to an appreciation of the domestic currency and how the monetary
authority (MA) might intervene to partly o¤set such an appreciation (per-
haps because, for reasons left out of the model, the MA fears that a large
appreciation might worsen the trade balance).
Consider the model of Section 2 with the only modi…cation that prefer-
ences are now given by:
Z 1
[u(ct ) + t v(mt )]e t dt; (114)
0

where should be thought of as a money demand shock. In the context

of this model:

(a) Consider the case of ‡exible exchange rates (with = 0). Suppose
that just before t = 0, the economy is in a stationary equilibrium with
a constant . At t = 0, there is unanticipated and permanent increase
in . Solve for the non-intervention case (i.e., a “pure ‡oating”).
Explain the intuition behind the results.
(b) Solve for the extreme case of “full intervention” (i.e., the MA reacts
in such a way that it does not let the nominal exchange rate change).
Explain intuitively how this policy works.
(c) Consider an “intermediate case” in which the MA chooses to inter-
vene in the foreign exchange market (but allows some of the adjust-
ment to take place through the nominal exchange rate). In particular,
derive a “policy reaction function”that would tell the MA how much
to intervene as a function of the change in real money demand (which
the MA must take as given) and the targeted change in the nominal
exchange rate. [HINT: a) Think of small changes so that you can
use di¤erentiation to compute changes at t = 0. b) Think of the MA
as having an initial positive stock of international reserves and that
capital gains/losses reserves are not monetized; that is, there is some
non-monetary liability (call it N M ) that is adjusted.]

2. Demand shocks
This exercise shows that, as one would expect, the dichotomy between the
real and the monetary sectors is still valid when the path of consumption
4 2 An answer key is available from the author upon request.

45
 is not constant over time. To this e¤ect, consider the following variation
of the model in the text. Preferences are given by:
Z 1
[ t u(ct ) + v(mt )]e t dt;
0

where t is a preference shock. The rest of the model is unchanged. The

parameter t can be viewed as a demand shock. Suppose that the path
of t is as follows:
H
; 0 t < T;
t = L
t T;
H L
where > . In this context:

(a) Solve for the perfect foresight equilibrium path corresponding to pre-
determined exchange rates.
(b) Solve for the perfect foresight equilibrium path corresponding to ‡ex-
ible exchange rates and show that it coincides with the one you just
derived for predetermined exchange rates.

3. Increase in domestic credit

Consider the economy analyzed in the text and operating under predeter-
mined exchange rates. In this context, analyze the e¤ects of an unantic-
ipated and permanent increase in the stock of domestic credit at time 0.
Explain the intuition behind the results.

4. Equivalence between predetermined and ‡exible rates illustrated

with an anticipated increase in the level of the money supply
Consider the economy analyzed in the text operating under ‡exible ex-
change rates. Suppose that the rate of money growth is zero (i.e., = 0)
and that the level of the money supply follows the path given by:

M L; 0 t < T;
Mt =
MH t T;

where M L < M H .
Suppose that the money supply is constant up to time T , increases at T ,
and remains unchanged thereafter. In this context:

(a) Solve for the perfect foresight path of all relevant variables.
(b) Show that if the economy were operating under predetermined ex-
change rates and the Central Bank set the path of the nominal ex-
change rate that you obtained in (a) above, the same equilibrium
would obtain.

46
5. In‡ationary consequences of anticipated changes in policy
This exercise explores yet another important distinction between a prede-
termined and ‡exible exchange rate systems: the behavior of the in‡ation
rate in response to an anticipated changes in policy.
Consider the model of Section 2. Characterize the perfect foresight equi-
librium paths corresponding to the following cases:

(a) Under predetermined exchange rates, suppose that the rate of deval-
uation is zero between 0 and T and increases to " > 0 at t = T . Solve
for the path of all relevant variables.
(b) Under ‡exible exchange rates, suppose that the rate of money growth
is zero between 0 and T and increases to > 0 at time T . Solve for
the path of all relevant variables.
(c) How does the behavior of in‡ation di¤er? What is the intuition
behind the results?

47
References
[1] Arrau, Patricio, Jose de Gregorio, Carmen M. Reinhart, and Peter Wick-
ham. 1995. The demand for money in developing countries: Assessing the
role of …nancial innovation. Journal of Development Economics 46 (2):
317-340.
[2] Bahmani-Oskooee, Mohsen, and Hafez Rehman. 2005. Stability of the
money demand function in Asian developing countries. Applied Economics
37 (7): 773-792.

[3] Calvo, Guillermo, and Leonardo Leiderman. 1992. Optimal in‡ation tax
under precommitment: Theory and Evidence. American Economic Review
82 (1): 179-194.
[4] Carlstrom, Charles T., and Timothy S. Fuerst. 2001. Timing and real in-
determinacy in monetary models. Journal of Monetary Economics 47 (2):
285-298.
[5] Cooley, Thomas, and Gary Hansen. 1989. The in‡ation tax in real business
cycle models. American Economic Review 79: 733-748.
[6] Cooley, Thomas, and Gary Hansen. 1995. Money and the business cycle. In
Frontiers of Business Cycle Research, ed. Thomas Cooley, 175-221. Prince-
ton University Press.
[7] Drazen, Allan, and Elhanan Helpman. 1987. Stabilization with exchange
rate management. Quarterly Journal of Economics 102 (4): 835-856.
[8] Drazen, Allan, and Elhanan Helpman. 1990. In‡ationary consequences of
anticipated macroeconomic policies. Review of Economic Studies 57 (1):
147-164.
[9] Easterly, William, Paolo Mauro, and Klaus Schmidt-Hebbel. 1995. Money
demand and seigniorage-maximizing in‡ation. Journal of Money, Credit,
and Banking 27 (2): 583-603.
[10] Goldfeld, Stephen M., and Daniel E. Sichel. 1990. The demand for money.
In Handbook of Monetary Economics, Volume I, ed. Benjamin M. Friedman
and Frank H. Hahn. Elsevier Science Publishers.
[11] Hansen, Lars. 1982. Large sample properties of generalized method of mo-
ments. Econometrica 50 (4): 1029-1054.

[12] Helpman, Elhanan. 1981. An exploration in the theory of exchange-rate

regimes. Journal of Political Economy 89 (5): 865-890.
[13] Levy Yeyati, Eduardo, and Federico Sturzenegger. 2005. Classifying ex-
change rate regimes: Deeds vs. words. European Economic Review 49 (6):
1603-1635.

48
[14] Lucas, Robert E., Jr. 1982. Interest rates and currency prices in a two-
country world. Journal of Monetary Economics 10 (3): 335-359.
[15] Reinhart, Carmen and Kenneth Rogo¤. 2004. The modern history of ex-
change rate arrangements: A reinterpretation. Quarterly Journal of Eco-
nomics 119 (1): 1-48.

[16] Reinhart, Carmen and Carlos Vegh. 1995. Nominal interest rates, consump-
tion booms, and lack of credibility: A quantitative examination. Journal
of Development Economics 46 (2): 357-378.
[17] Sargent, Thomas J. and Neil Wallace. 1981. Some unpleasant monetarist
arithmetic. Quarterly Review. Federal Reserve Bank of Minneapolis. Fall:
1-17.
[18] Svensson, Lars. 1985. Money and asset prices in a cash-in-advance economy.
Journal of Political Economy 93 (5): 919-944.

49
 Box 1. Exchange rate regimes in practice
Section 2.5 analyzes the main characteristics of predetermined exchange
rates and ‡exible exchange rates regimes. For conceptual clarity, we studied
those regimes in their "pure" form: under predetermined exchange rates, the
monetary authority controls the path of the nominal exchange rate and, under
‡exible exchange rates, the monetary authority does not intervene at all in the
exchange market. In real life, however, such "pure" forms are hard to …nd (par-
ticularly pure ‡oaters) and, in fact, we observe a whole continuum of exchange
rate regimes in between these two extremes. In addition, there are exchange
rate regimes that in fact go beyond predetermined exchange rates, such as "full
dollarization" (i.e., no domestic currency). What follows is an exchange rate
taxonomy – based on the degree of exchange rate ‡exibility and the existence
of formal or informal commitments to exchange rate paths – reported in the
2008 IMF report on “De Facto Classi…cation of Exchange Rate Regimes and
Monetary Policy Framework.”43

No separate legal tender The currency of another country circulates as

the sole legal tender or the member belongs to a monetary or currency union in
which the same legal tender is shared by the members of the union. Adopting
such regimes implies the complete surrender of the monetary authorities’control
over domestic monetary policy. Examples: Ecuador, Panama, and El Salvador.

Currency board arrangements A monetary regime based on an explicit

legislative commitment to exchange domestic currency for a speci…ed foreign
currency at a …xed exchange rate, combined with restrictions on the issuing
authority to ensure the ful…llment of its legal obligation. This implies that do-
mestic currency will be issued only against foreign exchange and that it remains
fully backed by foreign assets, leaving little scope for discretionary monetary
policy and eliminating traditional Central Bank functions, such as monetary
control and lender-of-last-resort. Some ‡exibility may still be a¤orded, depend-
ing on how strict the banking rules of the currency board arrangement are.
Hong Kong, for instance, has had a currency board (with the exchange rate
…xed at 7.8 Honk Kong dollars per one U.S. dollars) since October 17, 1983.
Other examples: Estonia, Lithuania, and Djibouti.44

Conventional …xed peg arrangements The country pegs its currency

within margins of 1 percent or less vis-à-vis another currency; a cooperative
4 3 See http://www.imf.org/external/np/mfd/er/2008/eng/0408.htm. The description of the

di¤erent regimes is taken from the IMF report. Examples given are current as of April 31,
2008. This classi…cation system is based on members’ actual, de facto, arrangements as
identi…ed by IMF sta¤, which may di¤er from their o¢ cially announced arrangements (see
the discussion below on de jure versus de facto regimes).
4 4 Argentina’s Converbitility Plan (1991-2001) would be another recent and highly publicized

example of a currency board.

50
arrangement, such as the ERM II; or a basket of currencies, where the bas-
ket is formed from the currencies of major trading or …nancial partners and
weights re‡ect the geographical distribution of trade, services, or capital ‡ows.
The currency composites can also be standardized, as in the case of the SDR.
There is no commitment to keep the parity irrevocably. The exchange rate
may ‡uctuate within narrow margins of less than 1 percent around a central
rate – or the maximum and minimum value of the exchange rate may remain
within a narrow margin of 2 percent – for at least three months. The mone-
tary authority maintains the …xed parity through direct intervention (i.e., via
sale/purchase of foreign exchange in the market) or indirect intervention (e.g.,
via the use of interest rate policy, imposition of foreign exchange regulations,
exercise of moral suasion that constrains foreign exchange activity, or through
intervention by other public institutions). Flexibility of monetary policy, though
limited, is greater than in the case of exchange arrangements with no separate
legal tender and currency boards because traditional Central Banking functions
are still possible, and the monetary authority can adjust the level of the ex-
change rate, although relatively infrequently. Examples: Venezuela, Honduras
and Argentina.

Pegged exchange rates within horizontal bands The value of the

currency is maintained within certain margins of ‡uctuation of more than 1
percent around a …xed central rate or the margin between the maximum and
minimum value of the exchange rate exceeds 2 percent. As in the case of con-
ventional …xed pegs, reference may be made to a single currency, a cooperative
arrangement, or a currency composite. There is a limited degree of monetary
policy discretion, depending on the band width. Examples: Slovak Republic,
Syria, and Tonga.

Crawling pegs The currency is adjusted periodically in small amounts at

a …xed rate or in response to changes in selective quantitative indicators, such as
past in‡ation di¤erentials vis-à-vis major trading partners, di¤erentials between
the in‡ation target and expected in‡ation in major trading partners. The rate of
crawl can be set to adjust for measured in‡ation or other indicators (backward
looking), or set at a preannounced …xed rate and/or below the projected in‡ation
di¤erentials (forward looking). Maintaining a crawling peg imposes constraints
on monetary policy in a manner similar to a …xed peg system. Examples:
Bolivia, Botswana and China.

Managed ‡oating with no predetermined path for the exchange

rate The monetary authority attempts to in‡uence the exchange rate with-
out having a speci…c exchange rate path or target. Indicators for managing
the rate are broadly judgmental (e.g., balance of payments position, interna-
tional reserves, parallel market developments), and adjustments may not be
automatic. Intervention may be direct or indirect. Examples: Algeria, Ukraine
and Colombia.

51
 Independently ‡oating The exchange rate is market-determined, with
any o¢ cial foreign exchange market intervention aimed at moderating the rate
of change and preventing undue ‡uctuations in the exchange rate, rather than
at establishing a level for it. Examples: United States, Japan, Mexico and Euro
area.
When it comes to quantifying the e¤ects of di¤erent exchange rate regimes
on, among other things, growth and in‡ation, a major issue confronting em-
pirical researchers is whether to use de jure or de facto classi…cations of ex-
change rate regimes. Until recently, most papers followed de jure classi…cations,
typically based on the approach that the IMF followed until 1997 of asking
member countries to self-declare their arrangements. More recently, however,
Levy-Yeyati and Sturzenegger (2002) and Reinhart and Rogo¤ (2004) have pro-
vided de facto classi…cations (like the one above by the IMF). These authors
argue that de jure classi…cations usually fail to describe actual exchange rate
regimes and that the gap between de facto (what countries actually do) and
de jure (what countries say they do) classi…cations may be critical for under-
standing/quantifying a host of important issues. An example of this gap is the
so-called "fear of ‡oating" phenomenon analyzed in Calvo and Reinhart (2000),
who argue that, while many emerging markets claim to pursue ‡exible exchange
rate arrangements, many in fact intervene heavily to keep the nominal exchange
rate within certain ranges. Table 3 illustrates the gap between de jure and de
facto regimes by comparing the traditional de jure IMF classi…cation with the
de facto classi…cation provided by Reinhart and Rogo¤ (2004). According to
this table, 487 annual observations with a de jure ‡exible exchange rate are
reclassi…ed by Reinhart and Rogo¤ (2004) as a …xed exchange rate regime. Two
interesting cases are China, 1994-1997, and Mexico 1992-1994, both with a de
jure ‡exible exchange rate regime but classi…ed as …xed by Reinhart and Rogo¤
(2004).

52
 Box 2. What is the interest rate elasticity of
money demand?
As shown in equation (14), the theoretical framework derived in this chap-
ter models money demand as a positive function of consumption and a negative
function of the nominal interest rate. The interest-rate elasticity of money
demand is a particularly important parameter for both researchers and poli-
cymakers alike because it tells us the extent to which money demand will be
a¤ected by changes interest rates and in‡ation.
The estimation procedure typically used in …nding the empirical counterpart
of expression (14) starts by developing an equilibrium model in discrete time
similar to the one outlined in Appendix 7.4. Given the timing assumptions of
most models, the interest rate enters the money demand equation through the
expression i=(1 + i), as is the case in the end-of-period speci…cation captured
in equation (104). Arrau et al (1995) introduce money in a general equilibrium
model through a transaction cost technology that takes the form

1 mt
H(mt ; t ; ct ) = h ; t ;
c1t ct
where H(:) represents transactions costs per unit of consumption (in other
words, total transactions costs are ct H(:)) and h(:) is given by

mt 1 mt mt mt
h ; t =K t + log ;
ct ct ct t ct
where t is a technological parameter that captures …nancial innovation and is
a parameter that represents the degree of scale economies in transaction (when
= 1, H is a function of the ratio m=c, which is the most common theoretical
speci…cation).
Under this functional form, it is easy to check that the money demand
equation can be expressed as:

it
log(mt ) = log( t ) + log(ct ) : (115)
1 + it
The parameter is usually referred to as he interest rate semi-elasticity of money
demand or, more generally, the opportunity cost semi-elasticity of money. Arrau
et al (1995) estimate the money demand function for a sample of ten developing
countries using quarterly time series from the mid-70’s to the early 80’s and
using M1 as their measure of money balances. Since their paper focuses on the
role of …nancial innovation in money demand, they allow for a time varying
t in equation (115). They also run regressions using i instead of i=(1 + i)
as the opportunity cost of holding money . Panel A in Table 4 presents their
estimates for for the …ve countries in which they …nd a long-run (cointegrating)
relation between the dependent and independent variables. Their estimates vary
between -0.5 and -3 and are, for the most part, signi…cantly di¤erent from zero.
Building up on the importance of …nancial innovation when modeling money
demand, Reinhart and Vegh (1995) develop a general equilibrium framework in

53
which money is used because of its services in reducing transaction costs. Under
certain functional forms, they derive the following money demand equation:

it
log(mt ) = + 'log( t ) + log(ct ) + log (116)
1 + it
where t is a proxy for …nancial innovation. The authors simultaneously esti-
mated equation (116) and the model’s intertemporal optimality condition (the
standard Euler equation) applying Hansen’s (1982) generalized method of mo-
ments (GMM). Their database consists of quarterly time series for the 70’s and
80’s for Argentina, Chile and Uruguay. Panel B in Table 4 presents their esti-
mates for . Notice that in this case corresponds to the interest rate elasticity
of money and not to a semi-elasticity as in equation (115).45 Their estimates
are signi…cantly di¤erently from zero for all countries and not inconsistent with
the …ndings of Arrau et al (1995).
Some researchers have argued that many developing countries, particularly
high-in‡ation ones, have gone through periods of interest rate controls, which
limits the use of the interest rate when measuring the opportunity cost of holding
money. Easterly et al (1995), for example, use the in‡ation rate, t , instead of
the interest rate when constructing their opportunity cost variable, estimating
an equation of the form

mt t
log = + ; (117)
yt 1+ t

where yt is output and is a parameter that captures the possibility of a non-

linear relationship between money balances and the corresponding opportunity
cost. The authors use output instead of consumption as their scale variable
because this aggregate is less subject to measurement error in their particular
sample of countries. Panel C presents the results of their estimations for a panel
of eleven high in‡ation countries using annual data for the 1960-1990 period,
and using M1 as their money balances measure. The …rst two rows correspond
to equation (117) estimated in levels, while the second two rows correspond
to equation (117) estimated in …rst di¤erences due to the di¢ culty of …nding
a cointegrating relationship in the levels speci…cation. The authors …nd that
the coe¢ cient is positive and statistically signi…cant in the nonlinear versions
of both speci…cations, and therefore argue that in high-in‡ation countries the
opportunity cost semi-elasticity of money is not constant but rather increases
with in‡ation.
How large is the interest rate elasticity of money in the United States?
Goldfeld and Sichel (1990) explain that until de mid-1970s the behavior of money
demand in the U. S. was well explained by a simple partial adjustment model
of the following form:

log(mt ) = a0 + a1 log(yt ) + a2 log(mt 1) + a3 log( t ) + log (it ) + errort (118)

4 5 It can be easily checked that the interest-rate elasticity equals i times the semi-elasticity.

54
 Panel D shows the results of estimating (118) applying a Cochrane-Orcutt
procedure to quarterly data for the 1952-1986 period, taking M1 as the relevant
measure of money holdings and using two rates to represent i: the commercial
paper rate (RCP) and the commercial bank passbook rate (RCBP). In this
case, the value of the interest rate elasticity is much smaller than the semi-
elasticity estimates for developing countries presented in panels A though C. The
authors also run regressions for di¤erent time periods and di¤erent variants of
equation (118), obtaining low interest rate elasticity estimates in almost every
case. Estimates for other G-7 countries summarized in Goldfeld and Sichel
(1990) also point out to small numbers.
Taking into account the overall perform of their estimations, Goldfeld and
Sichel (1990) suggested the need for rethinking the conventional speci…cation,
and experimented with di¤erent alternatives like estimating the money demand
equation by maximum likelihood, using last day of quarter ‡ow of funds data for
M1, and amending the initial model by a bu¤er-stock component in the partial
adjustment equation. In every instance, the estimated elasticity is below 0.04.
We thus conclude that the low value of the estimated interest rate elasticity of
money in the U.S. provides some empirical support for modeling money through
a cash in advance constraint. Indeed, this has been the route followed by Cooley
and Hansen (1989, 1995) in their seminal work introducing money in a real
business cycles model.

55
 Figure 1. Permanent devaluation

A. Exchange rate B. Consumption

E Cc

0 time 0 time

C. Nominal interest rate D. Real money balances

i m

r m

0 time 0 time

E. Real domestic credit F. International reserves

d h

0 time 0 time
 Figure 2. Permanent increase in devaluation rate

A. Rate of devaluation B. Consumption

e Cc

0 time 0 time

C. Nominal interest rate D. Real money balances

i m

0 time time
0

E. Real domestic credit F. International reserves

d h

0 time 0 time
 Figure 3. Permanent increase in domestic credit

A. Domestic credit B. Consumption

D Cc

0 time 0 time

C. Real money balances D. International reserves

m h

mss

0 time 0 time

E. Real domestic credit F. Trade balance

.
d TB

0 time 0 time
 Figure 4. Permanent increase in rate of money growth

A. Rate of money growth B. Consumption

m Cc

0 time 0 time

C. Real money balances D. Nominal exchange rate

m Log(E)

0 time time
0

E. Rate of depreciation F. Nominal interest rate

e i

0 r

0 time 0 time
 Figure 5. Equivalence between predetermined and flexible exchange rates I

A. Rate of devaluation B. Nominal exchange rate

e log(E)

eL
H
e

eL
E0 eH

0 time 0 time
T T

C. Nominal interest rate D. Real money balances

i m

r+eH

r+eL

0 T time 0 T time

E. Nominal money supply F. Rate of money growth

log(M) m

eL

H
e

eH eL

0 T time 0 T time
 Figure 6. Equivalence between predetermined and flexible exchange rates II

A. Rate of money growth B. Nominal money supply

m log(M)

mL
H
m

mL mH
M0

0 time 0 time
T T

C. Real money balances D. Nominal interest rate

m i

m(mL)
i0

m(mH)
r+mL

0 T time 0 T time

E. Rate of depreciation F. Nominal exchange rate

e log(E)

mH
e0 eL

mL

0 T time 0 T time
 Figure 7. Negative money demand shock under predetermined exchange rates

A. Money shock parameter B. Consumption

g Cc

0 time time
T 0 T

C. Real money balances D. International reserves

m h

0 T time 0 T time

E. Rate of depreciation F. Nominal interest rate

e i

e r + e + p*

0 T time 0 T time
 Figure 8. Negative money demand shock under flexible exchange rates

A. Money shock parameter B. Consumption

g Cc

0 T time 0 time
T

C. Real money balances D. Rate of depreciation

m e

m(gH)
m0

e0
m(g ) L
m - p*

0 T time T time
0

E. Nominal exchange rate F. Nominal interest rate

log(E) i

i0
r+m

0 T time T time
0
 Table 1
Balance Sheet
E t H ∗t Mt
Dt
 Table 2.  Nominal anchors

Flexible Predetermined
E Set by policy Endogenous
H* Endogenous Set by policy
D Set by policy Set by policy
M Endogenous Set by policy
Set by policy
NM n/a Endogenous

Note: n/a means "not applicable"
 Table 3. De jure versus de facto exchange rate regimes

RR de facto classification
Fixed Flexible
IMF de jure Fixed 1363 366
classification Flexible 487 303

Source: IMF and Reinhart and Rogoff (2002). Data points are annual
observations for 132 developing countries for the period 1970-2001
 

                  Table  4.   Empirical estimates of the opportunity cost semi‐elasticity of money 
A.                    Arrau et al (1995). Country‐specific regressions* 
  Estimated β when the opportunity cost measure is: 
Country  i  i/(1+i) 
Argentina    ‐0.47 
(‐2.14) 
Brazil    ‐2.17 
(‐3.50) 
India  ‐2.83   
(‐1.98) 
Israel    ‐2.97 
(‐13.70) 
Korea  ‐2.97   
(‐1.14) 
B.                   Reinhart and Vegh (1995). Country‐specific regressions* 
Country  Estimated β** 
Argentina  ‐0.10 
(‐5.00) 
Chile  ‐0.09 
(‐2.25) 
Uruguay  ‐0.22 
(‐2.20) 
C.                              Easterly et. al. (1995). Panel regressions* 
Specification  Estimated β  Estimated γ 
Levels – linear (γ=1)  ‐1.42   
(‐11.46) 
Levels – Non‐linear  ‐1.53  1.59 
(‐10.04)  (6.78) 
First‐differences – linear (γ=1)  ‐0.74   
(‐6.53) 
First‐differences – Non‐linear  ‐0.92  2.20 
(‐4.15)  (4.06) 
D.                       Goldfled and Sichel (1990). Money Demand in the US* 
Interest Rate   Estimated β**
 
RCP  ‐0.013 
(5.2) 
RCBP  ‐0.003 
(0.9) 
* t‐values in parenthesis 
**Estimates correspond to the interest rate elasticity of money