EC321 : Monetary Economics

Topic 1 — Interest rates and monetary policy

Kevin Sheedy

LSE

Lent term 2014

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 1 / 57

Monetary policy: Interest rates or money supply?

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 2 / 57

Interest rates and monetary policy
Traditional monetary theory developed in terms of the quantity
of money determining the (nominal) interest rate and the price
level.
But the total quantity of money seems to play little role in many
central banks’ monetary policy strategies:
I either as an explicit target: central banks have shifted targets
from intermediate to ultimate goals (e.g. “inflation targeting”)
I or as an instrument: monetary policy seen as directly setting the
interest rate
I (there is the recent experience with quantitative easing, but
central banks seem to consider this a policy only for abnormal
times)
Why has this happened?
And how might we understand the determination of interest
rates and the price level without reference to the money supply?
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 3 / 57
Financial innovation and money demand

The efficacy of targeting the money supply as a means of achieving

price stability depends on a stable money demand function.
Financial innovation and deregulation seems to have significantly
increased the volatility of money demand.
The argument is that more financial instruments are now available
that are close substitutes for the usual functions of money.
Some examples:
credit cards and electronic payment systems
“sweep” accounts and removal of interest-rate restrictions on
current accounts reduces opportunity cost of holding money

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 4 / 57

Instability of money demand
‘We didn’t abandon the monetary aggregates, they
abandoned us.’ — Gerald Bouey (former governor of the
Bank of Canada)
e.g. U.S. velocity of money:

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 5 / 57

Operating target of monetary policy
Poole’s (1970) analysis:
Money supply as operating target with shocks to money demand:
i
Ms

i2

i1 M2d
M1d
M

Leads to undesirable fluctuations in interest rates.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 6 / 57

Operating target of monetary policy
Interest rate as operating target with shocks to money demand:
i
M1s M2s

i∗
M2d
M1d
M

Poole’s analysis suggests that central banks should use interest rates
as their operating targets if money demand shocks are large and
frequent.
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 7 / 57
Controlling the interest rate

Nominal interest rate is not a direct instrument of policy — it is a

market price.
Instruments of monetary policy:
1 Open-market operations
2 Standing facilities (e.g. discount rate)
3 Reserve requirements
Which to use?
Traditional analysis stresses open-market operations that change the
money supply (in particular, reserves) and thus indirectly affect
interest rates.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 8 / 57

Interest rates and reserves
No apparent relationship in the data (e.g. Australia):
Interest rate:

Reserves:

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 9 / 57

Choice of instrument

Changing reserve requirements:

I a blunt instrument
I efficacy depends on reserves not being remunerated (which is
inefficient)
Open-market operations:
I volatility in demand for reserves requires high-frequency
open-market operations, which may still fail to control
short-term interest rates precisely
Standing facilities:
I can be used to set bounds on the market interest rate and thus
guide it more accurately than open-market operations alone

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 10 / 57

Channel/corridor system

In recent years, many central banks have moved away from

implementing monetary policy mainly through open-market
operations.
The new systems are designed to work more automatically by making
greater use of standing facilities:
Deposit facility
Borrowing facility
This type of system is known as a channel (or corridor) system.
To understand how this operating procedure works, we build a simple
model of the interbank market.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 11 / 57

 A model of the interbank market

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 12 / 57

A model of the interbank market

Market participants: n banks

Banks can hold reserves (deposits at the central bank)
Reserves are used for settling transfers between banks arising
from the payments system (e.g. a customer of one bank writes a
cheque payable to the customer of another)
Banks are unsure of the transfers they will need to make
Banks can borrow reserves from one another in the interbank
market, but do not know how much reserves they will require at
the point they participate in the interbank market.
Banks can also acquire reserves through the repo (repurchase
agreement) market. The central bank can be a participant in
this market.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 13 / 57

The model
Notation:
Bj = initial holding of reserves by bank j (after previous debts
settled)
Ij = net borrowing in interbank market (negative for lending) by
bank j
Pj = net bond repos (sale and repurchases) by bank j
Tj = net payment bank j must make to other banks (uncertain)
Rj = balance of bank j’s account at end of time period
(uncertain because Tj is not known in advance)
Accounting:
R j = Bj + I j + P j − Tj

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 14 / 57

Interbank market and repo market
Interbank market:
Uncollateralized lending between banks (potentially risky)
Interest rate = i
Repo market:
Sell bond (e.g. Treasury bond) now and agree to repurchase (at
a higher price, reflecting the repo interest rate) later
Effectively a collateralized loan, so risk free if collateral is good.
Let us assume that banks are not thought likely to default on interbank
loans, so these are perceived to be risk free. We should then expect to see
similar interest rates in these markets.
Our model will suppose that the interbank interest rate is exactly the
same as the repo rate for simplicity.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 15 / 57

Central-bank standing facilities
1 Deposit facility: central bank offers an interest rate of id on
positive balances in banks’ accounts at the end of the time
period
2 Borrowing facility: central bank charges an interest rate of ib
on negative balances in banks’ accounts at the end of the time
period
I positive spread over deposit rate: ib > id
I equivalent to an offer to make loans at the end of period, with
the loan credited to the bank’s account
I in practice, collateral required — we shall assume banks have
enough collateral to use it
These are referred to as standing facilities because banks know the
facilities will be available for them to use as needed at known interest
rates (and without any restriction on the magnitude of their use).

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 16 / 57

Balance on account next period
Reserve balance of bank j at the beginning of next period (after
interbank loans repaid, repos mature, and standing facility rates
paid/levied):
(
id Rj if Rj ≥ 0
Bj0 = Rj − (1 + i)(Ij + Pj ) +
ib Rj if Rj < 0

Since Rj = Bj + Ij + Pj − Tj :
(
id (Bj + Ij + Pj − Tj ) if Tj ≤ Bj + Ij + Pj
Bj0 = Bj −i(Ij +Pj )−Tj +
ib (Bj + Ij + Pj − Tj ) if Tj > Bj + Ij + Pj

Transfer Tj is unknown until end of period, so Bj0 is uncertain when

banks choose Ij and Pj . Density function of Tj is f (Tj ) and
distribution function is F (Tj ). Assume E[Tj ] = 0.
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 17 / 57
Bank objective function
Assume risk-neutral banks. Aim to maximize expected next period
reserve balance E[Bj0 ]. Choice variables: interbank transaction = Ij ,
repo market transaction = Pj .

Z Bj +Ij +Pj
E[Bj0 ] = Bj − i(Ij + Pj ) + id (Bj + Ij + Pj − Tj )f (Tj )dTj
Tj →−∞
Z ∞
+ib (Bj + Ij + Pj − Tj )f (Tj )dTj
Tj =Bj +Ij +Pj

First-order conditions:
∂E[Bj0 ] ∂E[Bj0 ]
= 0 and =0
∂Ij ∂Pj

These lead to the same equation, so we only need to consider one.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 18 / 57
First-order condition
Note that:
Z Bj +Ij +Pj Z Bj +Ij +Pj
∂
(Bj + Ij + Pj − Tj )f (Tj )dTj = f (Tj )dTj
∂Ij Tj →−∞ Tj →−∞

+ ((Bj + Ij + Pj ) − (Bj + Ij + Pj )) f (Bj + Ij + Pj ) = F (Bj + Ij + Pj )

Similarly:
Z ∞
∂
(Bj + Ij + Pj − Tj )f (Tj )dTj = 1 − F (Bj + Ij + Pj )
∂Ij Tj =Bj +Ij +Pj
Thus, the first-order condition reduces to:
−i + id F (Bj + Ij + Pj ) + ib (1 − F (Bj + Ij + Pj )) = 0
Or equivalently (at margin, expected benefit = expected loss):
(ib − i)(1 − F (Bj + Ij + Pj )) = (i − id )F (Bj + Ij + Pj )
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 19 / 57
Optimal use of interbank and repo markets
The optimal demand for interbank and repo lending is characterized
by the equation:
ib − i
F (Bj + Ij + Pj ) =
ib − id

Since we know 0 ≤ F (Bj + Ij + Pj ) ≤ 1, the first-order condition

implies id ≤ i ≤ ib , so the interest rate i must lie in the channel
bounded by id and ib in equilibrium.
The distribution function F (·) is an increasing function, so there
must be a negative relationship between i and Ij + Pj .
Initial balance Bj is predetermined, so equation implies banks with
high Bj will have low or negative values of Ij + Pj , i.e. they will be
lenders in the interbank and/or repo markets.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 20 / 57

Aggregation and market clearing

Aggregate beginning-of-period balances: B = nj=1 Bj

P

Aggregate end-of-period balances: R = nj=1 Rj

P

Payments between banks cancel out: nj=1 Tj = 0

P

Interbank market transactions cancel out: nj=1 Ij = 0

P

Repo market Ptransactions net out to central bank’s open-market

n
operation: j=1 Pj = P, where P is (reverse) repos (bond purchases,
with resale agreement) by central bank
Since Rj = Bj − Tj + Ij + Pj , it follows that R = B + P in the
aggregate, so P is net injection of new reserves.
R
Since Bj + Ij + Pj is the same for all banks: Bj + Ij + Pj = n

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 21 / 57

Equilibrium interbank interest rate
The equilibrium interbank interest rate i (and also the repo rate) is
determined by the equation:

ib − i
 
R
F =
n ib − id

The central bank directly controls:

the deposit rate id
the borrowing rate ib
open-market operations P
These indirectly determine:
Total reserves at end of period R = B + P (with B being
reserves at beginning of period)
The interbank and repo rate i
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 22 / 57
Interbank market with channel system
i

Rs
ib
i∗
Rd
id

0 R

ib −i
Demand: The equation F Rn =


ib −id
, which implies a negative
relationship between i and R
Supply: The equation R = B + P (inelastic — assumed
open-market operations are at central bank’s discretion)
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 23 / 57
Comparative statics

Holding constant other policy instruments:

An increase in id “squashes” the demand curve from below :
interbank rate i rises
An increase in ib “stretches” the demand curve from above :
interbank rate i rises
An increase in total reserves R shifts the supply curve to the
right : interbank rate i falls
A simultaneous increase of id and ib by the same amount implies a
parallel upward shift of the demand curve.
If net payment Tj has a symmetric probability distribution then
setting R = 0 implies i will be exactly halfway between id and ib .

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 24 / 57

Changing interest rates
Normally we think that changing interest rates requires a carefully
calibrated open-market operation (need to have a good knowledge of
the money demand function).
Much easier with a channel system: Given R, id and ib , suppose the
interest rate is initially i, which satisfies F (R/n) = (ib − i)/(ib − id ).
Now suppose central bank would like to increase i by x, i.e. achieve
i 0 = i + x. Set new deposit and borrowing rates id0 = id + x and
ib0 = ib + x. Note that

ib0 − i 0 (ib + x) − (i + x) ib − i
 
R
0 0
= = =F
ib − id (ib + x) − (id + x) ib − id n

so no new open-market operation (R 0 = R) is required.

Intuition: reserve demand depends on interbank rate relative to
standing facility rates in channel system.
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 25 / 57
Success of channel-based system
U.S. (traditional system) — greater volatility around target than

interest rate in Australia (channel system)

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 26 / 57

How wide should the channel be?
A narrower channel has advantages:
Guarantee that interest rate fluctuations are smaller
Less need for “fine-tuning” open-market operations
But there can be disadvantages as well. Think about the special case
of a channel of zero width (id = ib ):
Trading in interbank market dries up
Central bank becomes intermediary for all borrowing and lending
between banks.
I This means the central bank incurs all the costs and
inconvenience of monitoring the credit-worthiness of banks
(since banks are borrowing only from the central bank, not from
one another)

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 27 / 57

Alternative systems
The framework developed here can be used to analyse alternative
systems of interest rate control:
1 The traditional system
I Zero deposit rate id = 0
I Fixed penalty rate ib
I Market interest rate i controlled by varying reserves through
open-market operations (shifting the R s curve)
2 The “floor” system
I Standing facility rates id and ib , as in channel system
I Market is saturated with reserves (R s is shifted so far to the
right that it intersects R d at, or very close to, i = id )
I Market interest rate controlled by varying deposit rate id
I Advantage: No need for “fine-tuning” open-market operations
I Disadvantage: Very little trading in interbank market, in
practice this illiquidity might make the market work less well
I A system like this has been implemented in New Zealand
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 28 / 57
Controlling the price level with interest rate policy

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 29 / 57

Monetary policy through setting interest rates

Now we have seen how the central bank can control the market
nominal interest rate on bonds, we now ask what interest rate should
be set to achieve the central bank’s objectives.
Strategy: Identify some objective and set the interest rate to be
consistent with that objective.
Turns out that this is not sufficient. Need to specify a policy
reaction function to the state of the economy.
Analyse this problem in the simplest possible case:
sole aim of central bank is an inflation target
economy facing no shocks
Looks trivial — but actually not...

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 30 / 57

A simple model: Assumptions
Time periods: 0, 1, 2, . . . (nothing special about period 0 — just
the current time period)
Representative household: ct = consumption at time t
Utility = u(c0 ) + βu(c1 ) + β 2 u(c2 ) + · · ·
β = discount factor (0 < β < 1, representing impatience), u(c)
is an increasing and concave function (representing preference
for consumption smoothing)
Household income yt derives from an endowment of goods
(equivalent to inelastic labour supply)
Cashless economy (money is only a unit of account, not a
medium of exchange, nor a store of wealth)
Only asset is bond denominated in monetary units of account
(nominal bond): Bt = quantity of bonds held by household, it =
(nominal) interest rate
Price of goods in terms of money = Pt
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 31 / 57
Household budget constraint
At the beginning of period t, the household receives payment
(1 + it−1 )Bt−1 (in money terms) from bonds acquired in the past.
Household endowment generates income Pt yt in money terms.
Household consumption costs Pt ct
Household bond purchases (saving) = Bt
Budget constraint at time t:

Pt ct + Bt = Pt yt + (1 + it−1 )Bt−1

Can solve this to find consumption ct :

Bt−1 Bt
ct = yt + (1 + it−1 ) −
Pt Pt

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 32 / 57

Utility maximization
Substitute budget constraint into utility function:
 
B−1 B0
Utility = u y0 + (1 + i−1 ) −
P0 P
  0
B0 B1
+βu y1 + (1 + i0 ) − + ···
P1 P1

Differentiate with respect to bonds B0 and set equal to zero:

1 0 1
− u (c0 ) + β(1 + i0 ) u 0 (c1 ) = 0
P0 P1
Rearranging gives us the “consumption Euler equation”:
P0 0
u 0 (c0 ) = β(1 + i0 ) u (c1 )
P1

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 33 / 57

Euler equation
Differentiating utility with respect to B1 , B2 , etc. shows that the
Euler equation must hold for all t ≥ 0 to maximize utility:
Pt 0
u 0 (ct ) = β(1 + it ) u (ct+1 )
Pt+1
Writing this in terms of inflation πt+1 = (Pt+1 − Pt )/Pt :
 
0 1 + it
u (ct ) = β u 0 (ct+1 )
1 + πt+1
The Fisher equation defines the implied real return rt :
1 + it
1 + rt = hence u 0 (ct ) = β(1 + rt )u 0 (ct+1 )
1 + πt+1
u 0 (ct )
and MRSct ,ct+1 = = 1 + rt
βu 0 (ct+1 )
(i.e. equate marginal rate of substitution to price of period t + 1
consumption relative to period t consumption)
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 34 / 57
Equilibrium: market clearing
Goods market:
ct = yt
Consumption is only source of demand (no investment or government
spending here).

Bond market (although we can forget about this because of Walras’

law):
Bt = 0
All households identical (representative household assumption).
Cannot have a positive or negative demand for bonds in equilibrium:
suppose all households want to sell bonds (borrow) — who
would buy the bonds?
suppose all households want to buy bonds (save) — whom
would they buy the bonds from?
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 35 / 57
Simplifying the problem
Let us now make the model very simple:
Real income yt is constant over time: yt = y
The central bank’s only goal is an unchanging inflation target
π ∗ : πt∗ = π ∗
Constant income and goods-market clearing imply
ct = yt = y = yt+1 = ct+1 . It follows that u 0 (ct ) = u 0 (ct+1 ), so the
Euler equation reduces to
1
1 + rt =
β
The equilibrium real return on bonds must be constant:
rt = (1/β) − 1. Note this is higher when households are more
impatient (lower β) and so have a greater desire to borrow.
Fisher equation:
(1 + it ) = (1 + r )(1 + πt+1 )
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 36 / 57
Graphical analysis of equilibrium
rt
Ys

r
Yd
y yt

Equivalent to the dynamic model from EC210 Macroeconomic

Principles with:
Inelastic output supply (income comes from fixed endowment)
Output demand comprising only consumption
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 37 / 57
Monetary policy

Central bank would like to use monetary policy to achieve its inflation
target, i.e. πt = π ∗ for all t
The Fisher equation tells us what nominal interest rate i ∗ is
consistent with this target:

i ∗ = (1 + r )(1 + π ∗ ) − 1

We will now consider what happens when the central bank pegs the
nominal interest rate at i ∗ :

it = i ∗ for all t

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 38 / 57

Inflation

Clearly, achieving the inflation target at all times is an equilibrium:

πt = π ∗ . This satisfies all the required equations.
However, is this the unique equilibrium?
In EC210 Macroeconomic Principles, you studied cases where there
can be more than one equilibrium, e.g.:
the Diamond-Dybvig banking model
the coordination-failure model of business cycles
It turns out that the interest-rate peg can also lead to multiple
equilibria.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 39 / 57

Equilibria
We know that the Fisher equation must hold for all t ≥ 0:

1 + it = (1 + r )(1 + πt+1 )

Given the interest-rate peg it = i ∗ :

1 + it = 1 + i ∗ = (1 + r )(1 + π ∗ )

Together:

(1 + r )(1 + πt+1 ) = (1 + r )(1 + π ∗ ) and thus πt+1 = π ∗

So we know: π1 = π2 = · · · = π ∗
In other words, we must have an equilibrium where inflation is
expected to be on target in every period after the current one.
What about π0 ? — Unfortunately, no equation pins this down!
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 40 / 57
Multiple equilibria
Starting from a given point in time (period 0), the equilibria for
inflation are:

π0 can take any value, π1 = π2 = · · · = π ∗

π1 , π2 , . . . are the future inflation rates expected in period 0.

That there are equilibria with π0 6= π ∗ is bad, but at first glance the
problem does not look too bad: once we are past period 0, inflation
must be on target.
Not quite! — we have used “period 0” only as a label for the
“current period” — at any point in time we can always conclude
that current inflation is not uniquely determined.
All that the interest rate peg guarantees is that people must expect
inflation to be on target in the future — permits random fluctuations
around the target.
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 41 / 57
Intuition
Why is future inflation (period 1 onwards) uniquely determined, but
not current inflation (period 0)?

Given that monetary policy pegs the nominal interest rate, the
Fisher equation implies that future inflation has a negative effect
on real interest rates, which matter for consumption decisions:

rt ≈ i ∗ − πt+1

Since there is a unique equilibrium for the real interest rate,

there is a unique equilibrium for the inflation rate expected in
the future
Period 0 inflation only affects the ex post real return of bonds
acquired in the past (period −1), which does not affect current
or future consumption decisions.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 42 / 57

Intuition, continued
πt+1
s
Y
d
Y

π∗

y yt

Real interest rate: rt ≈ i ∗ − πt+1

With an unchanging nominal interest rate, the expectation of
higher future inflation lowers the real interest rate
Desire to consume more now (t) and less in the future (t + 1)
Upward pressure on Pt and downward pressure on Pt+1
Expectations of future inflation πt+1 = (Pt+1 /Pt ) − 1 fall,
bringing expectations into line with target.
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 43 / 57
Resolving the problem: A “feedback” rule

The problem: Monetary policy is “passive” — although consistent

with inflation target, does not rule out other equilibria in which the
target is not met
The solution: An “active” monetary policy that responds to the
current state of the economy, in this case, to inflation.
This type of monetary policy (with feedback from inflation to interest
rates) is known as a Taylor rule.
The simplest case of the Taylor rule:
For every 1% that the current (gross) inflation rate 1 + πt exceeds
the inflation target, the (gross) nominal interest rate is increased by
α% above its average.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 44 / 57

Simple Taylor rule
Writing the simple Taylor rule mathematically:

(1 + it ) − (1 + i ∗ ) (1 + πt ) − (1 + π ∗ )
 
=α
1 + i∗ 1 + π∗

where α is a parameter (the strength of the response of interest rates

to inflation).
The interest-rate peg it = i ∗ is the special case α = 0.
For a small real interest rate r and inflation target π ∗ we have the
approximations i ∗ ≈ r + π ∗ and

(1 + it ) − (1 + i ∗ ) (1 + πt ) − (1 + π ∗ )
∗
≈ it − i ∗ and ∗
≈ πt − π ∗
1+i 1+π

The Taylor rule is approximately it ≈ i ∗ + α(πt − π ∗ )

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 45 / 57
Equilibria
What equilibria are possible under the Taylor rule? — solve by
combining Taylor rule and Fisher equation.
First, write the Taylor rule as
1 + it 1 + πt
∗
=α + (1 − α)
1+i 1 + π∗
The Fisher equation requires 1 + it = (1 + r )(1 + πt+1 ) and
1 + i ∗ = (1 + r )(1 + π ∗ ), hence:
(1 + r )(1 + πt+1 ) 1 + πt
∗
=α + (1 − α)
(1 + r )(1 + π ) 1 + π∗
Multiply both sides by 1 + π ∗ :
1 + πt+1 = α(1 + πt ) + (1 − α)(1 + π ∗ )
Simplifying yields:
πt+1 = απt + (1 − α)π ∗

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 46 / 57

Difference equation

Under the simple Taylor rule, any equilibrium for inflation must
satisfy the first-order linear difference equation

πt+1 = απt + (1 − α)π ∗

for all t ≥ 0.
When |α| < 1, the solution of the difference equation is stable and
converges to π ∗ as t → ∞ for any initial condition π0 .
When |α| > 1, the solution of the difference equation explodes as
t → ∞ for any initial condition π0 except π0 = π ∗ .

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 47 / 57

The case of |α| < 1

The difference equation is stable. Looks like a good thing — but

not!
Normally, we think of solving a difference equation like
πt+1 = απt + (1 − α)π ∗ starting from some given initial condition at
time 0 — which would make sense if the variable was the capital
stock where there is an initial level based on past investment
decisions (this type of variable is referred to as a state variable).
Here, the variable in question is inflation, which is not a state
variable! (inflation today is not constrained by past inflation)
Any inflation rate π0 can be part of an equilibrium in which inflation
is subsequently expected to return to target gradually.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 48 / 57

Graphical analysis of the case 0 < α < 1
Many stable equilibria — two examples starting from π0 & π00 :
πt+1

πt+1 = απt + (1 − α)π ∗

π∗

0 0 0 πt
π0 π1 π2 π ∗ π3 π2 π1 π00
45◦

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 49 / 57

The case of |α| > 1

Now, any π0 other than equal to π ∗ results in an unstable path for

inflation expectations.
Possible equilibria:
Inflation is always on target: π0 = π1 = π2 = · · · = π ∗
Either hyperinflation or hyperdeflation is expected in the future
Assuming we can rule out the cases of hyperinflation and
hyperdeflation, there is a unique equilibrium remaining in which both
current and future inflation is always on target.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 50 / 57

Graphical analysis of the case α > 1
All equilibria are unstable, except starting from π0 = π ∗
πt+1
πt+1 = απt + (1 − α)π ∗

π∗

0 0 0 πt
π2 π1 π0π ∗π0π1 π2
45◦

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 51 / 57

The “Taylor principle”

This analysis shows that it is not sufficient that the interest rate
responds to inflation to rule out multiple equilibria; it must do so
sufficiently aggressively.
The Taylor principle: The nominal interest rate must be increased
by more than the increase in inflation (α > 1)
This response is needed to invalidate what would otherwise be
self-fulfilling beliefs that inflation would deviate from target.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 52 / 57

Taylor rule and stabilization policy
Although this was not part of our analysis in this lecture, the Taylor
principle can also be important for stabilizing the economy in the
short run:
Suppose an increase in demand pushes up inflation
If there was no change in the nominal interest rate, the higher
inflation would reduce real interest rates
That would stimulate demand and create further inflationary
pressure
But if the Taylor principle is satisfied, nominal interest rates are
raised by more than inflation increases by
This raises the real interest rate, reducing demand, and thus
helping to control inflationary pressure
In this lecture, we have focused instead on the role of the Taylor
principle in providing a “nominal anchor” - a way of pinning down the
level of prices in terms of money.
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 53 / 57
Taylor rule: a description of U.S. monetary policy

When proposing the rule, Taylor (1993) argued that it was a good
description of (the outcome) of U.S. monetary policy decisions.
Taylor’s proposal featured an interest rate response to the output gap
as well as inflation:

it = r̄ + 2% + 1.5(πt − 2%) + 0.5(yt − ȳt )

where yt is (log) output and ȳt is potential output, r̄ is the long-run

real interest rate, and the inflation target is implicitly assumed to be
2%.
Taylor chose a proportional response of the interest rate to inflation
of 1.5, and 0.5 for the response to the output gap.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 54 / 57

Comparing the Taylor rule and actual policy
decisions

Very different when Burns was chairman; quite similar for Greenspan.
K. Sheedy (LSE) EC321 Monetary Economics Topic 1 55 / 57
Evidence on the Taylor principle

Clarida, Gali & Gertler (2000) look at how the Taylor rule that best
fits the U.S. data has changed over time.
They estimate a Taylor rule of the form:

it = ρit−1 + (1 − ρ)(i ∗ + β(πte − π ∗ ) + γ(yt − ȳ ))

where πte is expected inflation as of time t, and yt − ȳ is the

output gap (deviation of actual from potential output).
The Taylor rule includes an interest-rate smoothing effect when
ρ > 0 — the actual interest rate adjusts slowly to what would
be implied by the standard Taylor rule.
Look at different sub-samples.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 56 / 57

Taylor principle

Taylor principle depends on β being above one.

True for post-Volcker period, but not prior to Volcker.
Interpretation: U.S. monetary policy in 70s did not satisfy Taylor
principle — hence inflation expectations not anchored.
May explain failure to control inflation around that time.

K. Sheedy (LSE) EC321 Monetary Economics Topic 1 57 / 57