Examiners commentaries 2015

Examiners commentary 2015

EC3115 Monetary Economics
Mock examination paper
Section A
Indicate whether the following statements are true or false, or uncertain and give a short
explanation. Points are only given for a well-reasoned answer.
Question 1
A positive relationship between the nominal interest rate and anticipated inflation
implies that the rate of growth of money has no effect on the real economy.
The statement is uncertain. The question is about the superneutrality of money, where the
steady-state values of real variables (such as consumption, capital) are independent of the rate of
growth of nominal money or inflation rate. This is a property of some of the general equilibrium
models with money, for example, the Sidrauski money-in-the-utility function (MIU) model or the
deterministic cash-in-advance (CIA) model. With superneutrality, the relation between the
nominal interest rate and anticipated inflation must, via the Fisher equation, be one-for-one.
Perhaps the most important reason superneutrality will fail in actual economies is the presence of
taxes that are not indexed to inflation; for instance, being levied on nominal capital gains (rather
than real capital gains) so that effective tax rates will depend on the inflation rate, generating
real effects on capital accumulation and consumption as inflation varies.

Question 2
The marginal efficiency of general taxation has no bearing on the optimal amount of
seigniorage revenue.
The statement is false. Assuming that all taxation is distortionary in some sense, minimizing
overall distortions means collecting at least some seigniorage. For a given amount of tax revenue,
the optimal level of seigniorage is linked to the problem of determining the optimal level of all
taxes. A change in the marginal distortionary cost of general taxation per unit of revenue will
affect the optimal mix of all taxes, including seigniorage.

Question 3
Fiat money has no commodity backing, but it does have intrinsic value.
The statement is false. Commodity money (such as gold or silver) has intrinsic value. Fiat
money (such as government-issued banknotes) is money that has physical substance but no
intrinsic value. Its use is established by custom and practice. The value of fiat money comes from
its usefulness in facilitating exchange with others who accept it in payment, even though it is
intrinsically worth nothing.

EC3115 Monetary Economics

Question 4
An inverted yield curve can signal that financial markets expect a recession.
The statement is true. Because the yield curve contains information about future expected
interest rates, it has the capacity to help forecast future output fluctuations (as well as inflation).
Rising interest rates are associated with economic booms and falling interest rates with
recessions. When the yield curve is either flat or downward sloping (i.e. inverted), it suggests
that future short-term interest rates are expected to fall and, therefore, that the economy is more
likely to enter a recession.

Question 5
The policy ineffectiveness proposition continues to hold even when the government
has better information than private agents.
The statement is false. Where the government has informational and time advantages, changes in
policy that are systematic can have real, short-run effects because they respond to shocks that
the public cannot perceive. This is true for the Lucas misperceptions model. An excellent
mathematical illustration is provided by question 17 of the 2010 examination paper. The
example shows how the government can reduce output fluctuations when it has an informational
advantage over the private sector.

Question 6
An inflation bias exists even when policymakers target the market-clearing level of
output.
The statement is false. Typically, (time consistent) discretionary monetary policy leads to an
inflationary bias in monetary policy that depends on the policymakers objective function.
Suppose the policymaker has a loss function which is quadratic in both the inflation rate and the
output gap, meaning that it tries to keep inflation and the output gap as close as possible to the
target levels in the loss function. Discretionary monetary policy can lead to an inflation bias if
the targeted level of output in the loss function is above the natural level of output. The
ultimate result of such a policy is a higher rate of inflation than desired but no gain in output.
However, if the targeted level of output in the loss function is the same as the natural level of
output then no inflation bias occurs.

Question 7
Relative purchasing power parity states that higher domestic inflation relative to
abroad will be reflected in an appreciation of the domestic currency.
The statement is false. The question has relative purchasing power parity the wrong way round:
higher domestic inflation compared to abroad means the exchange rate depreciates by the
inflation differential.

Question 8
Assuming risk neutrality and rational expectations, market efficiency implies that
forward exchange rates are positively correlated with future spot exchange rates.
The statement is uncertain. Applying rational expectations and risk neutrality leads to an
equation where the log of the future spot exchange rate is equal to the log of the forward

Examiners commentaries 2015

exchange rate today plus a random error term with mean zero. Hence, while we should expect a
positive correlation, the test of efficiency would be whether the correlation between forward
exchange rates and future spot exchange rates is unity.

Section B
Question 9
Consider a central bank that would like to minimize the expected squared deviation
of inflation from a target of zero, that is, minimize the loss function E[ 2 ].
Assume that the central bank is able to use the instruments of monetary policy to
control aggregate demand y perfectly. The relationship between inflation and
aggregate demand is known to be:
= ay z
where z is a supply shock and a is a positive coefficient.
(a) (4 marks) Assume first that the central bank observes both the supply shock z
and the coefficient a perfectly, so there is no uncertainty. Find the choice of
demand y that minimizes the loss function of the central bank.
(b) (8 marks) Now assume that while the central bank knows the coefficient a, it is
uncertain about the shock z. The central bank forms a rational expectation of z
using the information it has available, where z = E[z] is its estimate of z, and
= z z is the error it makes (which is known to have standard deviation ).
Find the choice of demand y that minimizes the expected loss of the central
bank. Comment on your answer in comparison to part (a).
(c) (8 marks) Now assume the central bank observes the shock z perfectly, but does
not know the coefficient a. Its rationally formed estimate of a is a
= E[a], and
the error it makes is = a a
(which is known to have standard deviation ).
Find the choice of demand y that minimizes the expected loss of the central
bank. Comment on your answer in comparison to parts (a) and (b).
(a) The loss function of the central bank is the expected squared inflation rate (consistent with
a symmetric target for zero inflation for a central bank with no other objectives):
L = E[ 2 ]
Inflation is given by = ay z, where y is aggregate demand (under the control of the
central bank), where a is a positive coefficient and z is an exogenous shock. The loss
function in terms of y, a, and z is:
L = E[(ay z)2 ]
If there is no uncertainty about either z or a then the central bank can perfectly predict the
inflation rate that will result from its choice of y:
L = (ay z)2
The optimal choice of y minimizes the loss function. The first-order condition is:
L
= 2a(ay z) = 0
y
This requires:
ay z = 0

EC3115 Monetary Economics

and hence:

z
a
If there is a positive supply shock z (one that reduces inflationary pressure), the central
bank should boost demand, and reduce demand when there is a negative supply shock. The
magnitude of the required response of monetary policy to the shock is inversely related to
the coefficient a (a larger value of a makes inflation more sensitive to demand).
y=

(b) The central bank now does not know z, but has a rationally formed estimate z = E[z] using
all available information. The central bank continues to know the value of a. In this case,
the inflation rate that results from a choice of y is uncertain. The optimal policy is the one
that minimizes the expected value of the loss function, for which the first-order condition is:



L
=
E[(ay z)2 ] = E
(ay z)2 = E[2a(ay z)] = 2aE[ay z] = 0
y
y
y
which is obtained by moving the differentiation operator inside the expectation operator.
The central bank knows a, and knows its own choice of y, so this equation becomes:
0 = E[ay z] = E[ay] E[z] = ay z
implying that the optimal value of y is:
y=

z
a

The answer is very similar to that obtained in part (a). The only difference is that the
unknown value of the shock z is replaced by the central banks best estimate, but the
magnitude of the response to the expected shock is the same. The standard deviation of the
error has no effect on the optimal policy, so it does not matter how confident the central
bank is of its estimate. This is the certainty equivalence principle: the policymaker should
respond to the best estimate of an unknown variable in exactly the same way as if the value
were known with certainty. This is a general result that holds whenever the loss function is
quadratic, the equations describing the economy are linear with the unknown variable in
question (the supply shock here) entering additively.
(c) Now assume that it is the coefficient a that is not known by the central bank (the shock z is
once again known with certainty). Let a
= E[a] denote the central banks best estimate of
the unknown a using all the information it has available rationally. The error the central
bank makes is = a a
, which is known to have standard deviation . The optimal choice
of y is once again found by minimizing the expected loss function. Following the first few
steps in part (b), the first-order condition is:


L

=
E[(ay z)2 ] = E
(ay z)2 = E[2a(ay z)] = 2E[a(ay z)] = 0
y
y
y
Unlike part (b), a is now a random variable and cannot be moved outside the expectation
operator. The first-order condition can be manipulated as follows:
0 = E[a2 y az] = E[a2 y] E[az] = E[a2 ]y zE[a]
which is justified because neither y nor z is a random variable (z is known to the
policymaker, and y is chosen by the policymaker). Since a = a
+ (where a
= E[a]), it
follows that:
E[a2 ] = E[(
a + )2 ] = E[
a2 + 2
a + 2 ] = a
2 + 2
aE[] + E[2 ] = a
2 + 0 + Var() = a
2 + 2
and therefore the first-order condition is:
0 = (
a2 + 2 )y a
z
The optimal value of y is therefore:
y=

a
z
a
2 + 2

Examiners commentaries 2015

and by dividing numerator and denominator by a

:
y=

z
2
a
+ a

This solution is very different from what was found in parts (a) and (b). Compared to part
(a), the unknown value of a is not simply replaced by the best available estimate a
. The
presence of the term 2 /
a means that the certainty equivalence principle is not valid in this
example. The extent of uncertainty about a determines how much the central bank should
optimally respond to the supply shock z. Since 2 /
a is positive in the presence of
uncertainty about a, the magnitude of the policy response to the supply shock is smaller
with uncertainty about a. In other words, policy is more cautious than it would be if
information were better (part (a) is the special case where = 0).
The analysis in part (b) is an example of shock uncertainty or additive uncertainty, for
which certainty equivalence holds. Part (c) is an example of parameter uncertainty or
multiplicative uncertainty. Intuitively, with parameter uncertainty, and hence uncertainty
about the monetary policy transmission mechanism, the more the policymaker reacts to
shocks, the greater is the uncertainty about the ultimate objectives the policymaker cares
about. This suggests a policy that reacts less, and thus creates less uncertainty about its
outcome. With shock uncertainty, the economys response to policy is known perfectly and
it is simply a matter of matching this to the estimated shocks.

Question 10
Consider the expectations theory of the term structure. The short-term (one year)
interest rate is denoted by it , the yield on a two-year bond is jt (referred to as the
long-term yield), and st = jt it is the yield spread (also referred to as the slope of
the yield curve). The expectations theory predicts that the long-term yield is:
jt =

1
1
it + Et it+1
2
2

where Et denotes expectations conditional on information available at time t.

(a) (4 marks) Explain the argument that justifies the relationship between jt , it ,
and Et it+1 in the equation above.
Suppose monetary policy is conducted so that the short-term interest rate is
it = i + t , where i is the average interest rate and t is a mean-zero i.i.d. shock with
variance 2 .
(b) (4 marks) Derive a formula in terms of i and t for the long-term yield jt and
the yield spread st implied by the expectations theory.
(c) (8 marks) For each of the following, find the prediction of the expectations
theory and evaluate it with reference to empirical evidence on the yield curve:
i. The average slope of the yield curve: E[st ]
ii. The covariance between the short and long yields: Cov(it , jt )
iii. The covariance between the short rate and the yield spread: Cov(it , st )
iv. The relative variance of the long and short yields: Var(jt )/Var(it )
(d) (4 marks) Now suppose the relationship between the short and long yields
includes a risk premium term t :
jt =

1
1
it + Et it+1 + t
2
2

In cases where the implications in part (c) do not fit the data, explain what
assumptions on the risk premium t are needed to match the data.

EC3115 Monetary Economics

(a) The argument is that the total expected return from holding a short-term bond between t
and t + 1 and then rolling over the investment into another short-term bond between t + 1
and t + 2 is (approximately) it + Et it+1 , where it is the yield on the short-term bond at time
t (and the return between t and t + 1), and it+1 is the yield on the short-term bond at time
t + 1. The total return from holding the long-term bond between t and t + 2 is
(approximately) 2jt . If investors are risk neutral, they will hold the bond with the greatest
expected return. Since both maturities must be held in equilibrium, the yields must adjust
to make investors indifferent between them. This requires that the expected returns are the
same:
1
1
2jt = it + Et it+1 jt = it + Et it+1
2
2
(b) Monetary policy is assumed to be conducted in a way that means there are unpredictable
fluctuations in the short-term interest rate it around its mean value i, that is, it = i + t ,
where t is an i.i.d. shock. The implied long-term yield is:
jt =

1
1
1
1
t
1
(i + t ) + Et [i + t+1 ] = i + t + i = i +
2
2
2
2
2
2

since Et t+1 = 0 (the shock t+1 cannot be predicted given what is known at time t). The
implied yield spread st = jt it is:
t
t
st = i +
(i + t ) =
2
2
(c) i. The prediction for the average of the yield curve slope st is:
h i
1
t
= E[t ] = 0
E[st ] = E
2
2
since t has (unconditional) mean zero. The average yield-curve slope is predicted to be
zero. This prediction is not consistent with the data, where the finding is that the yield
curve is upward sloping on average.
ii. The prediction for the covariance of between the short- and long-term yields it and jt is:

  1
t 
1
2
t
Cov(it , jt ) = Cov i + t , i +
= Cov t ,
= Cov(t , t ) = Var(t ) =
2
2
2
2
2
which uses that the covariance of a random variable with itself is the same as its variance.
The theory is seen to predict that short- and long-term yields have a positive covariance,
which means that they are positively correlated. This is consistent with the empirical
finding that yields of different maturities tend to move together over time on average.
iii. The prediction for the covariance between the yield-curve slope st and the short rate it is:


t 
t 
1
1
2
Cov(it , st ) = Cov i + t ,
= Cov t ,
= Cov(t , t ) = Var(t ) =
2
2
2
2
2
The theory predicts that the short rate and the yield-curve slope have a negative
covariance, which means that they are negatively correlated. This is consistent with the
empirical finding that the yield curve tends to slope upwards when short-term rates are
unusually low and is more likely to have a negative slope when short-term yields are very
high.
iv. The variance of the short-term rate is Var(it ) = Var(i + t ) = Var(t ) = 2 . The variance
of the long-term rate is:

   1 2
t 
2
t
Var(jt ) = Var i +
= Var
=
Var(t ) =
2
2
2
4
The prediction for the relative variance of the long- and short-term yields is therefore:
2

Var(jt )
1
= 42 =
Var(it )

so the long-term rate should have a lower variance than the short-term rate (it should
fluctuate less on average). This is not consistent with empirical evidence that suggests
long-term yields have a similar volatility to short-term yields.

Examiners commentaries 2015

(d) The predictions of the expectations theory for the average slope of the yield curve and the
relative variance of short- and long-term yields failed to match the data. Now consider
version of the expectations theory equation augmented with a risk premium t :
jt =

1
1
it + Et it+1 + t
2
2

Following the same steps as before, the long-term yield is:

jt =

1
1
t
(i + t ) + Et [i + t+1 ] + t = i +
+ t
2
2
2

and the yield spread st = jt it is:

st = i +

t
t
+ t (i + t ) = + t
2
2

The average yield-curve slope is:

h
i
t
E[st ] = E + t = E[t ]
2
Hence, the risk premium t on long-term bonds must be positive on average to explain why
the yield curve is upward sloping on average. The variance of the short-term rate
Var(it ) = 2 is the same as before. The variance of the long-term rate is:


 1

t
t
+ t = Var
+ t = Var(t ) + Var(t ) + Cov(t , t )
Var(jt ) = Var i +
2
2
4
Since it = i + t , the covariance Cov(t , t ) is equal to Cov(it , t ). The prediction for the
relative variance is:
2
+ Var(t ) + Cov(it , t )
Var(jt )
= 4
Var(it )
2
If the risk premium t is sufficiently volatile, and if the covariance between the risk premium
t and the short-term rate is not too negative, it is possible that the long-term yield can
have a similar volatility to the short-term yield.

Question 11
Consider an economy where the output gap yt and inflation t are related according
to the Lucas supply function:
yt = (t Et1 t )
where Et1 denotes expectations conditional on information available at time t 1,
and is a positive constant. The equation for aggregate demand is:
yt = mt pt
where pt is the price level (the inflation rate is t = pt pt1 ), and mt is the money
supply, which is set exogenously by the central bank.
(a) (5 marks) By noting that Et1 yt = 0 and Et1 pt = Et1 mt , show how the solutions
below for the output gap and the price level are derived:
yt =

(mt Et1 mt )
1+

and

pt =

mt +
Et1 mt
1+
1+

(b) (5 marks) Assume that monetary policy is set according to mt = mt1 + t , where
t is a mean-zero i.i.d. shock. Find solutions for the output gap yt and inflation
t in terms of current and past values of the shock t . Use your answer to
deduce the relationship between t , yt , and yt1 .

EC3115 Monetary Economics

(c) (5 marks) Now assume that monetary policy is set according to

mt mt1 = mt1 mt2 + t , where t is a mean-zero i.i.d. shock. Find the
relationship between t , t1 , and current and past values of yt .
(d) (5 marks) How do your answers to (b) and (c) help to understand the unstable
statistical relationship between inflation and the output gap? Relate your
answer to the different specifications of Phillips curve equations that have been
suggested by economists, and use this example to explain what is meant by the
Lucas Critique.
(a) Taking expectations of the supply equation conditional on period t 1 information:
Et1 yt = Et1 [(t Et1 t )] = (Et1 [t ] Et1 [Et1 [t ]] = (Et1 t Et1 t ) = 0
which uses Et1 [Et1 t ] = Et1 [t ], confirming that Et1 yt = 0. Now take expectations of
the demand equation conditional on information available at time t 1 and use Et1 yt = 0:
0 = Et1 yt = Et1 [mt pt ] = Et1 mt Et1 pt
and hence Et1 pt = Et1 mt . Next, observe that:
t Et1 t = (pt pt1 ) Et1 [pt pt1 ] = (pt Et1 pt ) (pt1 pt1 ) = pt Et1 pt
since pt1 = Et1 pt1 (there is no uncertainty about pt1 at time t 1). Using this and the
earlier finding Et1 pt1 = Et1 mt , the supply equation becomes:
yt = (pt Et1 pt ) = (mt yt Et1 mt )
where the second equality uses the demand equation to deduce pt = mt yt . It follows that:
yt + yt = (mt Et1 mt )

yt =

(mt Et1 mt )
1+

Using pt = mt yt again and the solution for yt :




(mt Et1 mt ) = 1
mt +
Et1 mt =
mt +
Et1 mt
pt = mt
1+
1+
1+
1+
1+
which confirms the solution for pt .
(b) Now assume that monetary policy is described by the equation mt = mt1 + t , where t is
an i.i.d. shock. This means that the money supply follows a random walk (the best forecast
of any future level of the money supply is its current level):
Et1 mt = Et1 [mt1 + t ] = Et1 mt1 + Et1 t = mt1 + 0 = mt1
since Et1 t . This expression can be substituted into the solutions for yt and pt derived in
part (a):

yt =
(mt mt1 ) =
t
1+
1+
and:
1

pt =
mt +
mt1
1+
1+
The solution for inflation t = pt pt1 is obtained by subtracting the equation above at
time t 1 from the time t equation:
t =

(mt mt1 ) +
(mt1 mt2 ) =
t +
t1
1+
1+
1+
1+

Since the solution for yt implies t = ((1 + )/)yt , the relationship between inflation t and
the output gap yt is:
t =

1 1+
1+
1
yt +
yt1 = yt + yt1
1+
1+

This implies that the level of inflation is positive related to the current and recent past
values of the output gap.

Examiners commentaries 2015

(c) Now suppose that monetary policy is described by the equation

mt mt1 = mt1 mt2 + t . This says that the growth rate mt mt1 of the money
supply follows a random walk. That is, the best prediction of the future growth rate is the
current growth rate. To find the solution, note that:
Et1 mt = Et1 [mt1 + (mt1 mt2 ) + t ] = mt1 + (mt1 mt2 )
since Et1 t = 0. Substituting this back into the formula for yt from part (a) and using
mt = mt1 + (mt1 mt2 + t :
yt =

((mt1 + (mt1 mt2 ) + t ) (mt1 + (mt1 mt2 ))) =

t
1+
1+

and the formula for pt :

pt =

1
(mt1 + (mt1 mt2 ) + t ) +
(mt1 + (mt1 mt2 ))
1+
1+

Since 1/(1 + ) + /(1 + ) = 1, the formula for pt simplifies to:

pt = mt1 + (mt1 mt2 ) +

1
t
1+

Inflation t = pt pt1 is obtained by subtracting the equation for the price level at time
t 1 from the time t equation:
t = mt1 + (mt1 mt2 ) +

1
1
t mt2 (mt2 mt3 )
t1
1+
1+

and by grouping together similar terms:

t = (mt1 mt2 ) + ((mt1 mt2 ) (mt2 mt3 )) +

1
(t t1 )
1+

Noting that (mt1 mt2 ) (mt2 mt3 ) = t1 according to the description of

monetary policy, it follows that:
t = (mt1 mt2 ) + t1 +

1
1

(t t1 ) = (mt1 mt2 ) +
t +
t1
1+
1+
1+

Subtracting the equivalent of this equation for past inflation t1 leads to:
t t1 = (mt1 mt2 ) (mt2 mt3 ) +

(t t1 ) +
(t1 t2 )
1+
1+

and using (mt1 mt2 ) (mt2 mt3 ) = t1 again:

t = t1 + t1 +

1
(t t1 ) +
(t1 t2 )
1+
1+

and simplifying:
1

t + 2
t1
t2
1+
1+
1+
The solution for yt implies t = ((1 + )/)yt , so t is related to t1 , yt , yt1 , and yt2 as
follows:
1
t = t1 + yt + 2yt1 yt2

This equation implies that a combination of current and past output gaps is related to the
difference between current and past inflation (the change in the inflation rate t t1 ).
t = t1 +

(d) The Phillips curve obtained in part (b) is the closest to the original Phillips curve
specification in relating the level of inflation to a measure of real economic activity. The
Phillips curve obtained in part (b) is the accelerationist version of the Phillips curve (the
change in the inflation rate is related to a measure of real economic activity) this is the
type of Phillips curve that underlies the concept of NAIRU (non-accelerating inflation rate
of unemployment). It is also related to the expectations-augmented Phillips curves that

EC3115 Monetary Economics

were suggested before the rational expectations revolution, with past inflation as a proxy for
inflation expectations (which implicitly assumes a simple form of adaptive expectations:
people simply extrapolate from past inflation when forming expectations of subsequent
inflation).
As this question shows, neither of these specifications is the true Phillips curve in this
example, in the sense of a Phillips curve equation that is structural (does not depend on
how monetary policy is conducted). The correct specification of the Phillips curve in this
example is:
1
t = Et1 t + yt

which states that the output gap is related to the deviation of actual inflation from the
rationally expected rate of inflation. This Phillips curve shifts with any change in what rate
of inflation is rationally expected. The link between inflation expectations and observable
variables is not stable and depends on the conduct of monetary policy, as the examples in
parts (b) and (c) show.
It follows that the form of the statistical relationship between inflation and the output gap
under a particular monetary policy is not structural and can not generally be exploited by
policymakers. For example, the Phillips curve in part (b) suggests there is a relationship
between the rate of inflation and the output gap, and a policymaker might therefore
conclude that output can be raised permanently by having a permanently higher rate of
inflation. However, the statistical relationship does not allow the policymaker to reach that
conclusion. Intuitively, the monetary policy in part (b) does not ever imply a permanent
change in the inflation rate (the inflation rate is a stationary variable under the policy), so
the data provided by past experience do not contain examples of the consequences of
permanent changes in inflation. And once policy changes to one where there are permanent
changes in the inflation rate (the monetary policy in part (c) is one way this could happen),
the statistical relationship changes to suggest that permanently rising inflation is needed to
sustain higher output permanently. It is easy to show that this statistical relationship is not
exploitable, either. These examples provide an illustration of the Lucas critique of
policymakers using statistical, not structural, relationships to evaluate the consequences of
different policies.

Question 12
Suppose that households around the world consume both tradable (T ) and
non-tradable (N ) goods with prices PT and PN in terms of domestic currency, and
prices PT and PN in terms of foreign currency. The consumer price levels P and P
in terms of domestic and foreign currency are:
P = PT1 PN

and

P = PT 1 PN

where is the weight on non-tradable goods (0 < < 1). The law of one price holds
for tradable goods:
PT = SPT
where S is the nominal exchange rate (the domestic-currency price of a unit of
foreign currency).
(a) (5 marks) Derive an expression for the real exchange rate s = SP /P in terms of
the relative prices of non-tradable and tradable goods in the domestic and
foreign economies.
(b) (2 marks) Under what condition will purchasing power parity (PPP) hold?
Suppose that tradable and non-tradable goods are produced in both economies
using labour, which is mobile between sectors, but not between countries. The
domestic nominal wage is W and the foreign nominal wage is W . Labour
productivity in tradables is denoted by AT in the domestic economy and AT in the

10

Examiners commentaries 2015

foreign economy. Both economies share the same level of labour productivity in
non-tradable goods (AN = AN ). All prices are determined in perfectly competitive
markets, hence PT = W/AT for example.
(c) (7 marks) Find an expression for the real exchange rate in terms of the relative
productivity in tradables AT /AT of the domestic economy.
(d) (6 marks) What is the empirical finding known as the Balassa-Samuelson
effect ? How can your answer to part (c) provide an explanation of this
phenomenon?
(a) Substituting the formulas for the domestic and foreign price indices P and P into the
definition of the real exchange rate s = SP /P :
 
SPT (PN /PT )
PN /PT
SPT 1 PN
=
=
s=
1

PT (PN /PT )
PN /PT
P T PN
which uses the law of one price PT = SPT for tradable goods.
(b) Purchasing power parity requires that the cost of a basket of goods is the same in all
countries after correcting for currency differences, that is, P = SP , or equivalently, that the
real exchange rate s is equal to one. The expression derived in part (a) shows that PPP will
hold if and only if:
P
PN
= N
PT
PT
which states that the relative prices of non-tradable goods and tradable goods must be the
same in both countries. However, in the absence of trade for non-tradable goods, there is no
force that automatically brings these relative prices into line across countries.
(c) Given labour productivities AT and AN for tradable and non-tradable goods in the
domestic economy and nominal wages W in both sectors, the prices that would prevail with
perfectly competitive markets (price equals marginal cost) are:
PT =

W
AT

and PN =

W
AN

and similarly in the foreign economy where productivities are AT and AN (non-tradables
productivity does not differ across countries) and nominal wages W :
PT =

W
AT

and PN =

W
AN

The relative prices of non-tradable goods in terms of tradables in the two countries are thus:
PN
W/AN
AT
=
=
PT
W/AT
AN

and

W /AN
A
PN
=
= T

PT
W /AT
AN

It follows that the ratio of the relative prices across countries is:
PN /PT
A /AN
A
= T
= T
PN /PT
AT /AN
AT
and using the expression for the real exchange rate from part (a):
 
AT
s=
AT
(d) The Balassa-Samuelson effect is the empirical regularity that richer countries tend to have
currencies with a stronger real value than suggested by purchasing power parity (the value
of e is lower in richer countries and higher in poorer countries). If the main difference
between richer and poorer countries is that the former have higher levels of productivity in
tradable goods (AT is higher than AT ) then the analysis in part (c) predicts that e will be
lower. Higher productivity in tradables (with no difference in non-tradables) will also mean
that the country has a higher level of real income. This is consistent with the relationship
between per-capita income and the real exchange rate found in the data.

11

EC3115 Monetary Economics

Question 13
Consider a small open economy with perfect capital mobility. Prices are sticky in
the short run, but fully flexible in the long run. Consumption demand depends
positively on disposable income. Investment demand and government spending are
exogenous. Net exports are assumed to increase with the competitiveness of the
domestic economy and decrease with income. Demand for real money balances
depends positively on income and negatively on the nominal interest rate. The
nominal interest rate satisfies uncovered interest parity (UIP).
(a) (5 marks) Analyse the consequences of an unexpected permanent increase in
the money supply using the AA-DD model. What happens to output and the
nominal exchange rate in the short run? In the long run?
(b) (5 marks) Define what is meant by the term exchange-rate overshooting.
Present examples showing that both overshooting and undershooting are
possible in the AA-DD model.
(c) (5 marks) Now assume that money demand is independent of income and
depends only on the nominal interest rate. How does this assumption change
the AA-DD diagram? Following an increase in the money supply, does the
exchange rate overshoot or undershoot, or is it still uncertain?
(d) (5 marks) Assume again that money demand depends on both real income and
the nominal interest rate, and now also assume that investment spending is
negatively affected by the interest rate. How does this alternative assumption
change the AA-DD diagram? Taking a case from part (b) where exchange-rate
overshooting occurred, will there now be more or less overshooting with the
alternative assumption?
(a) A permanent increase in the money supply has two effects on the AA curve (which
represents the combinations of output Y and nominal exchange rates S consistent with
money-market and foreign-exchange market equilibrium). First, in the long run, an
expansion of the money supply will raise the price level and the exchange rate
proportionately. This is because long-run output is supply determined, and the nominal
interest rate will be pinned down by the foreign interest rate once no further changes in the
exchange rate are expected. Since the demand for real money balances depends on real
income and the nominal interest rate, the long-run price level must be proportional to the
money supply. Furthermore, given the level of competitiveness (real exchange rate)
consistent with demand being equal to the supply-determined level of output, a rise in the
price level requires a proportional depreciation of the nominal exchange rate to reach this
long-run equilibrium level of competitiveness.
In the short run, the permanent increase in the money supply therefore leads to expectations
that the domestic currency will have a lower value in the future. Given the nominal interest
rate consistent with money-market equilibrium, foreign-exchange market requires an
immediate fall in the value of the domestic currency (to be consistent with uncovered
interest parity) otherwise investors would be expecting a capital loss. This shifts the AA
curve upwards from AA0 to AA2 (note that the nominal exchange rate S is defined as the
domestic price of foreign currency, so a rise in S is a depreciation of the domestic currency).
The second effect on the AA curve results from an increase in the supply of real money
balances M/P in the short run owing to the stickiness of prices P . For a given level of
output Y , money-market equilibrium entails a lower domestic nominal interest rate to bring
money demand in line with supply. A lower nominal interest rate reduces the return on
domestic assets relative to foreign assets, and since investors must be willing to hold
domestic assets in equilibrium, it is necessary (given the expected level of the nominal
exchange rate in the future) for the domestic currency to experience an immediate loss of
value against foreign currency. This ensures that there is an expected appreciation of the
domestic currency between the short run and the long run that compensates for the lower
interest rate. In the diagram, this means that the AA curve shifts up further from AA2 to
AA1 .

12

Examiners commentaries 2015

There is no shift of the DD curve in the short run because prices are sticky. The new
short-run equilibrium is at the intersection between DD0 and AA1 and results in a rise in
output to Y1 above the long-run level of output Y and an immediate depreciation of the
nominal exchange rate (moving from S0 to S1 ).
S
DD2
DD0
S1
S2

AA1

S0

AA2
AA0
Y

Y1

Since output is above Y , domestic prices will rise in the long run. The rise in prices has two
effects. First, it completely reverses the second shift of the AA curve that resulted from the
increase in real money balances. With prices rising in proportion with the money supply,
real money balances will return to their original level in the long run, so the AA curve shifts
back from AA1 to AA2 . Second, the rise in prices reduces the competitiveness of domestic
goods at any given level of the nominal exchange rate, resulting in a reduction in demand.
This shifts the DD curve to the left until it is returned to a position consistent with output
Y in the long run, namely for prices to rise in line with the money supply and the exchange
rate, restoring competitiveness (the real exchange rate) to its original level. The DD curve
therefore shifts from DD0 to DD2 . The long-run level of output is back at Y , but the
exchange rate has still depreciated relative to its starting point (S1 is above S0 ).
(b) Exchange-rate overshooting occurs when the short-run response of the nominal exchange
rate to some shock is greater than its response in the long run. By choosing the gradients of
the AA and DD curves appropriately, it can be seen that there are examples where the
short-run exchange rate S1 rises above S0 more than the long-run exchange rate S2
(overshooting), and examples where the short-run exchange rate S1 rises by less than the
long-run exchange rate S2 (undershooting).
Undershooting

Overshooting
S

S
DD2
DD0

S1
S2

DD2
AA1

S0

AA2

S2
S1

DD0

S0

AA1

AA0
Y

Y1

AA2
AA0
Y1

(c) If the demand for money does not depend on income then this affects the nature of the AA
curve. Now, since changes in income do not shift the money demand curve, the interest rate
required for money-market equilibrium is the same for all levels of income. Given this
interest rate, uncovered interest parity implies the same exchange rate for all levels of
income. This argument demonstrates that the AA curve is horizontal in this case. However,
it continues to shift in the ways determined earlier when the money supply increases.

13

EC3115 Monetary Economics

With a horizontal AA curve, the movement of the exchange rate in equilibrium is the same
as the vertical shift of the AA curve. With two shifts upwards of the AA curve following a
permanent increase in the money supply, but only one remaining in the long run, the
exchange rate is now certain to overshoot its long-run value in the short run.

S
DD2
DD0
S1

AA1

S2

AA2

S0

AA0

Y1

(d) It is now assumed that investment demand depends negatively on the (real) interest rate.
Since investment is part of aggregate demand, this change will affect the DD curve, but not
the AA curve. The DD curve is drawn on axes with output and the nominal exchange rate,
and the positive gradient results from higher values of the exchange rate (depreciations)
raising net exports. Given the expected future value of the exchange rate, the uncovered
interest parity condition implies that higher values of the current exchange rate are
associated with lower interest rates (the greater the depreciation now, the larger the
expected appreciation in the future). These lower interest rates stimulate investment, which
also boosts demand in addition to the effect of the exchange rate on net exports. A given
rise in the exchange rate now has two effects on demand, so the increase in output is larger
than before, implying that the DD curve is flatter than the usual case.
The argument above is implicitly holding inflation expectations constant in going from the
nominal interest rate determined by uncovered interest parity to the real interest rate that
affects investment. The permanent increase in the money supply will create expectations of
inflation equal to the percentage increase in the money supply because prices do not change
in the short run but will adjust fully in the long run. That would imply a lower real interest
rate when the money supply increases permanently. Note that there is also an effect of the
higher level of the future exchange rate on the nominal interest rate according to uncovered
interest parity. The nominal interest rate must rise by the percentage increase in the level of
the future nominal exchange rate. Since that is equal to the percentage increase in the
money supply, the rise in the nominal interest rate brought about by this effect is equal to
the increase in expected inflation. Therefore, the net effect on the real interest rate is zero,
leaving only the effect discussed in the paragraph above. The only change is to the gradient
of the DD curve: there are no additional shifts to take account of.
Since the DD curve is flatter than the usual case, the change in the exchange rate in the
short run is smaller given that all the shifts of the AA curve are unaffected. Given that the
vertical shift of the DD curve in the long run is the same as usual (the size of the price
adjustment in the long run is still in proportion to the increase in the money supply), the
long-run equilibrium exchange rate is the same as before. It follows that exchange-rate
overshooting is now smaller or the exchange rate actually undershoots. Investment demand
depending on the interest rate therefore reduces the likelihood of observing exchange-rate
overshooting.

14

Examiners commentaries 2015

S
DD2
DD0

DD02
DD00

S1
S10
S2

AA1
S0
AA2
AA0
Y

Y1

Y10

15