Lecture 2 Monetary policy, a little bit

Introduction

Two topics:
▶ Interest rate policy in a cashless economy
⋆ a modern version of the classical dichotomy
⋆ one can talk about policy without the quantity theory
⋆ policy goal: not real effect but stabilize price
▶ An old style macro model
⋆ money is in the model
⋆ policy cannot be systematically stimulative as old style models purport
Interest rate policy in a cashless economy

Outline
▶ real economy in the Arrow-Debreu setup
▶ translate the Arrow-Debreu prices into nominal terms without the
quantity theory
▶ interest rate policy that leads to determinacy
Real economy
The agent’s optimal choice problem

max ∑ β t u(ct )
t≥0

subject to
∑ pt ct = ∑ pt ωt ,
t≥0 t≥0

where pt is the date-t good’s price in the date-0 good’s unit (i.e., the
date-0 good is numeraire andp0 = 1), and ct is the agent’s
consumption of the date-t good
An equilibrium is a sequence of the prices {pt }t≥0 such that the
optimal choice of the agent of the consumption the date-t good is
equal to ωt , all t
We shall have more formal treatment of an infinite-lived agent
decision problem in the future course
Real economy

Here, we know in equilibrium, the consumption of date-t good must

be ωt . But the prices must be right to ensure that the agent does not
choose something different. The condition for price to be right is the
familiar first order condition
pt+1 β u ′ (ωt+1 )
= ; (1)
pt u ′ (ωt )

▶ if the agent cuts the consumption of the date-t good by εt , then he

has unspent wealth pt εt in the date-0 good unit
▶ with this unspent wealth, he can increase his consumption of the
date-t + 1 good by pt εt /pt +1
▶ in order for the agent to not cut the consumption of the date-t good, it
must be that
εt u ′ (ωt ) = (pt εt /pt +1 )β u ′ (ωt +1 ),
which is (1)
Real economy

Let u(c) = lnc. Then (1) becomes

β ωt
pt+1 = pt . (2)
ωt+1

Further let ωt = ω all t. Then the equilibrium is

{pt }t≥0 = {1, β , β 2 , ...}
Translation

There is a central bank

▶ its liability is called money; the unit of money is dollar
▶ if you hold one dollar in whatever forms, then you can claim one dollar
from CB in the form you like (e.g., cash)
▶ central bank issues bonds
⋆ one unit of bonds issued t takes one dollar from you but becomes
(1 + it ) dollars at t + 1
⋆ it is set by the central bank
Translation
Now instead of the date-0 good, money is the numeraire
Moreover, instead of the Arrow-Debru style of trade, the agent trades
sequentially, i.e., each t he sells and buys the date-t good on the
date-t market
The agent’s optimal choice problem
max ∑ β t u(ct )
t≥0

subject to
Mt+1 + Bt+1 + Pt ct = Pt ωt + Mt + (1 + it−1 )Bt all t
where Pt is the nominal price date-t good in the date-t market,
good’s unit, Mt is the amount of money the agent holds at the start of t,
Bt is the amount of bonds the agent holds at the start of t, and
M0 = B0 = 0.
An equilibrium is a sequence of the prices {Pt }t≥0 such that the
optimal choice of the agent of the consumption the date-t good is
equal to ωt , all t
Translation
Again, we know in equilibrium, the consumption of date-t good must
be ωt . But the prices must be right to ensure that the agent does not
choose something different. The condition for price to be right is

Pt (1 + it ) ′
u ′ (ωt ) = β Et u (ωt+1 ); (3)
Pt+1

▶ the expectation term Et appears because Pt +1 may not be determinate

(but the agent knows its objective distribution and so can form the
expectation based on the information available in date t)
▶ if the agent cuts the consumption of the date-t good by εt , then he
has Pt εt more dollars in date t
▶ he can buy bonds this amount of dollar and increase his consumption
of the date-t + 1 good by (1 + it )Pt εt /pt +1
▶ in order for the agent to not cut the consumption of the date-t good, it
must be that

εt u ′ (ωt ) = Et ((1 + it )Pt εt /Pt +1 )β u ′ (ωt +1 )

Translation

Let u(c) = lnc and ωt = ω all t. Then (3) becomes

1 1
1 + it = [Et ]−1 , (4)
β 1 + πt+1

where 1 + πt+1 = Pt+1 /Pt is the gross inflation rate.

If inflation is fixed at zero, then
1
1 + it = . (5)
β
Translation

But fixing the nominal interest rate by (5) can only fix

1
Et
1 + πt+1

at unity but cannot fixed πt+1 at 0

▶ there are many objective random variables xt +1 satisfying
1
Et =1 (6)
1 + xt +1
▶ when the agent picks such a random variable xt +1 at t and expects
that the realization of xt +1 is the inflation at t + 1, his expectation is
rational
The Taylor rule
Now suppose that when the agent picks a random xt+1 to form his
inflation expectation, all realizations of xt+1 are around zero so (3)
can be approximated by the familiar Fisher equation

it = r + Et πt+1 , (7)

where r = 1/β − 1
The Taylor rule
it = φ πt + ρ. (8)
▶ The Taylor principle: coefficient φ > 1
▶ ρ a parameter
It follows from the Fisher equation and the Taylor rule that

πt = φ −1 [Et πt+1 + (r − ρ)] (9)

so
πt = ∑ φ −(s+1) (r − ρ) (10)
s≥0
Policy ineffectiveness in an old style model

Efforts have been made to build models in which

▶ policy in general and monetary policy in specific are effective in the
short run
Those models were popular and influential
But they have been abandoned
▶ Reasons? To be discussed here
Three equations
y : log aggregate output
y : log price level
m: log money stock
i: one-period nominal rate of interest
IS equation

yt = a0 + a1 (it − πte |t−1 ) + v1t , a1 < 0, (11)

LM equation

mt − pt = c0 + c1 yt + c2 it + v2t , c2 < 0 < c1 , (12)

Friedman-Phelps natural-rate aggregate supply function (AS)

yt = γ0 + γ1 (pt − pte |t−1 ) + γ2 yt−1 + ut , γ1 > 0, 1 > γ2 ≥ 0. (13)

Three equations

πte |t−1 : date-t-1 expectation of private agents regarding inflation

between t and t+1
pte |t−1 : date-t-1 expectation regarding the date-t price level
v1t , v2t , and ut are i.i.d. random variables
▶ unsystematic and unpredictable factors in economy
AS: output supply affected by price only when price differs from its
expected value
Rational expectations

Rational expectations:
e
xt+j |t−1 = Et−1 xt+j

for j ≥ 0,
▶ Et−1 xt +j is mathematical expectation of xt +j computed according to
⋆ equations of the model
⋆ values of all variables in the model realized in periods t-1 and earlier
RE

With RE, (11) becomes

yt = a0 + a1 [it − Et−1 (pt+1 − pt )] + v1t , a1 < 0, (14)

And (13) becomes

yt = γ0 + γ1 (pt − Et−1 pt ) + γ2 yt−1 + ut , γ1 > 0, 1 > γ2 ≥ 0. (15)

Policy

Monetary authority follows a rule

▶ relates target value of m or i to past values of variables in the model
(e.g., yt−1 , pt−2 )
A money-supply rule targets m; suppose the rule is

mt = µ0 + µ1 mt−1 + µ2 yt−1 + et , (16)

▶ et is i.i.d., mean zero, constant variance: policy error

▶ et independent of structural disturbances v1t , v2t , and ut .
Aggregate Demand Function

Eliminate it from IS (14) and LM(12) to get aggregate demand

function

yt = β0 + β1 (mt − pt ) + β2 Et−1 (pt+1 − pt ) + vt , (17)

where

β0 = (a0 c2 − a1 c0 )/(a1 c1 + c2 ),
β1 = a1 /(a1 c1 + c2 ) > 0,
β2 = −a1 c2 /(a1 c1 + c2 ) > 0,
vt = (c2 v2t − a1 v1t )/(a1 c1 + c2 ).
Equilibrium

Let yt in (17) be equated to yt in (15);

1
pt = [(β0 − γ0 ) + β1 mt + γ1 Et−1 pt − γ2 yt−1 (18)
γ1 + β1
+β2 Et−1 (pt+1 − pt ) + vt − ut ]

▶ this determines “equilibrium” price

Policy ineffectiveness
Applying Et−1 (.) to (18),

1
Et−1 pt = [(β0 − γ0 ) + β1 Et−1 mt + γ1 Et−1 Et−1 pt − γ2 yt−1
γ1 + β1
+β2 Et−1 Et−1 (pt+1 − pt ) + Et−1 (vt − ut )];

simplifying,
1
Et−1 pt = [(β0 − γ0 ) + β1 Et−1 mt + γ1 Et−1 pt − γ2 yt−1
γ1 + β1
+β2 Et−1 (pt+1 − pt )];

and subtracting,
1
pt − Et−1 pt = [β1 (mt − Et−1 mt ) + (vt − ut )]. (19)
γ1 + β1
Policy ineffectiveness
Applying Et−1 (.) to (16),

Et−1 mt = µ0 + µ1 mt−1 + µ2 yt−1 + Et−1 et ;

simplifying,
Et−1 mt = µ0 + µ1 mt−1 + µ2 yt−1 ;
and subtracting,
mt − Et−1 mt = e t . (20)
Substituting (20) into (19),

1
pt − Et−1 pt = [β1 (mt − Et−1 mt ) + vt − ut ] (21)
γ1 + β1
1
= [β1 et + vt − ut ].
γ1 + β1
Policy ineffectiveness

Substituting (21) into AS (15),

γ1 vt + β1 ut + γ1 β1 et
yt = γ0 + γ2 yt−1 + . (22)
γ1 + β1

In (22), equilibrium yt influenced only by

▶ disturbances vt , ut , et and
▶ non-policy parameters γ0 , γ1 , γ2 , β1 = a1 /(a1 c1 + c2 )
▶ NOT by µs in policy rule (16)
The Lucas Critique
Combining policy rule (16) and equilibrium output function (22) (to
eliminate et ),

yt = φ0 + φ1 yt−1 + φ2 mt + φ3 mt−1 + ξt , (23)

where

φ = γ1 β1 (γ1 + β1 ),
φ0 = γ0 − µ0 φ , φ1 = γ1 − µ1 φ , φ2 = φ , φ3 = −µ1 φ ,
ξt = φ [(ut /γ1 ) + (vt /β1 )].

Suppose monetary authority follows policy rule (16) in history.

Econometricians can estimate φ0 , φ1 , φ2 , and φ3 accurately
▶ φ2 > 0
▶ So increasing µs in policy rule (16) can increase output?
The Lucas Critique

The φ parameters in (23) are estimated when µs are fixed

When µs change, φ will change