Monetary Economics Notes

Nicola Viegi1

2010

1 University of Pretoria - School of Economics

Contents

1 New Keynesian Models 1

1.1 Readings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Basic New Keynesian Model . . . . . . . . . . . . . . . . . . . 2
1.2.1 Consumer Problem . . . . . . . . . . . . . . . . . . . . 2
1.2.2 The Firm . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Optimal Monetary Policy 9

2.0.3 Optimal Policy Problem . . . . . . . . . . . . . . . . . 9
2.0.4 Discretionary Solution . . . . . . . . . . . . . . . . . . 10
2.0.5 Commitment - Timeless Perspective . . . . . . . . . . . 10
2.0.6 Policy Inertia . . . . . . . . . . . . . . . . . . . . . . . 11

v
Chapter 1

New Keynesian Models

Modern New Keynesian models are the standard tool of modern macroeco-
nomic modelling for policy analysis. These models bridge the gap between
the methodology of the Real Business Cycle tradition and practical policy
evaluation. Introducing price stickiness and imperfect competition in the ba-
sic RBC model they provide an useful tool to analyse response of economies
to nominal and real shock.

1.1 Readings
Two books provide a complete overview of a very large literature. They are:

Woodford, Michael (2003): Interest and Prices, Princeton University

Press.

Gali, Jordi (2008) Monetary Policy, In‡ation and the Business Cycle,
Princeton University Press

The Gali’s book provide a very useful discussion of the literature at the end
of each chapter. The objective of the following notes is just to clarify some of
the maths passages necessary to follow the argument and are not substitute
to a careful reading of the Gali book.

1
2 CHAPTER 1 NEW KEYNESIAN MODELS

1.2 Basic New Keynesian Model

1.2.1 Consumer Problem
" #
X
1 1
Ct+i 1+
Nt+i
i
Et (1.1)
i=0
1 1+
Z 1 1 1

Ct = cjt dj (1.2)
0

Z 1 1
1

Pt = p1jt dj (1.3)
0

Pt Ct + Bt = Wt Nt + (1 + rt ) Bt 1 + t (1.4)
1) Optimal Allocation of Consumption Expenditure
Z 1
min pjt cjt dj (1.5)
cij 0
subject to
Z 1 1 1

Ct = cjt dj (1.6)
0

Lagrangian
Z " Z #
1 1 1 1

Lt = pjt cjt dj + Ct cjt dj (1.7)

0 0

FOC
" Z 1 #
1
@Lt 1 1 1
= pjt + cjt dj cjt =0 (1.8)
@cjt 0

Noting that
Z 1 1
1
1 1
cjt dj = Ct (1.9)
0

Rewrite (1.8) as
1.2 BASIC NEW KEYNESIAN MODEL 3

1 1
pjt Ct cjt = 0 (1.10)
1 1
pjt = Ct cjt (1.11)
1 1
pjt cjt = Ct (1.12)
1 1
cjt = Ct (1.13)
pjt

cjt = Ct (1.14)
pjt
Or, as usually expressed in the literature

pjt
cjt = Ct (1.15)

Substituting (1.15) in (1.2) we get:

"Z # 1 Z
1 1
pjt 1 1 1

Ct = Ct dj = pjt 1 dj Ct (1.16)
0 0

and solving for the Lagrange Multiplier / we get

Z 1 1

1 = pjt 1 dj (1.17)
0
Z 1 1

= pjt 1 dj (1.18)
0
Z 1 1
1

1
= pjt dj Pt (1.19)
0

Substituting this back in FOC (1.8), we get:

pjt
cjt = Ct (1.20)
Pt
2) Optimal Dynamic Consumption/Leisure Decision
4 CHAPTER 1 NEW KEYNESIAN MODELS

In real terms, the consumer problem can be written as:

" #
X1
C 1
t+i N 1+
t+i
i
Et (1.21)
i=0
1 1+
Bt Wt Bt 1 t
Ct + = Nt + (1 + rt ) + (1.22)
Pt Pt Pt Pt
Lagrangian
" #
X
1 1
Ct+i 1+
Nt+i
i
L = Et + (1.23)
i=0
1 1+
X1
Wt+i Bt+i 1 t+i Bt+i
i
Et t+i Nt+i + (1 + rt+i ) + Ct+i (1.24)
i=0
Pt+i Pt+i Pt+i Pt+i
FOC

@L
= Ct + t+i = 0 (1.25)
@Ct
@L Wt
= Nt + t =0 (1.26)
@Nt Pt
@L 1 1
= t + Et t+1 (1 + rt+1 ) =0 (1.27)
@Bt Pt Pt+i
Rearranging and using condition (1.25) to eliminate the Lagrange multiplier,
we get:

Wt
Nt = Ct (1.28)
Pt
Pt
Ct = Et (1 + rt+1 ) Ct+1 (1.29)
Pt+i
As shown before, this implies the following log linear relationships (which we
will use later) on

wt pt = c t + nt (1.30)
1
ct = Et fct+1 g fit Et t=1 ln g (1.31)
1.2 BASIC NEW KEYNESIAN MODEL 5

1.2.2 The Firm

Firms are pro…t maximisers but they fact three constraints:

Production Function linear in labour input (the simplest possible - there

is no capital - just looking at the short run properties of the model)

cjt = Zt Njt (1.32)

a downward sloping demand curve

pjt
cjt = Ct (1.33)
Pt+i

nominal inertia like in Calvo (1983).- in each period (1 ) are ran-

domly chosen to set their prices

Real Total Cost, Average Cost, Marginal Cost

Wt Wt
TC = Nt = cjt (1.34)
Pt Z t Pt
Wt
AC = M C = (1.35)
Zt Pt
Productivity shocks a¤ect marginal cost of the …rm
Firm pricing problem

X
1
pjt Wt+j
i
max = Et i;t+i cjt+i cjt (1.36)
pjt+i
i=0
Pt+i Zt+i Pt+i
" #
X
1
pjt pjt pjt
i
= Et i;t+i Ct+i M Ct+i (1.37)
Ct+i
i=0
Pt+j Pt+i Pt+i
" #
X
1
pjt
1
pjt
i
= Et i;t+i M Ct+i Ct+i (1.38)
i=0
Pt+i Pt+i

FOC for the optimal price pt

6 CHAPTER 1 NEW KEYNESIAN MODELS

" #
X
1
1 pjt
1
1 pjt
i
Et i;t+i (1 ) + M Ct+i Ct+i(1.39)
= 0
i=0
pjt Pt+i pjt Pt+i
X
1
pjt 1 pjt
i
Et i;t+i (1 ) + M Ct+i Ct+i(1.40)
= 0
i=0
Pt+i pjt Pt+i

Flexible Price Equlibrium

pt
= M Ct (1.41)
Pt 1
Wt
1= (1.42)
1 Z t Pt

Wt 1 Nt
= Zt = (1.43)
Pt Ct
Log linearizing

Nt
ln aZt = ln (1.44)
Ct
taking total derivatives and evaluating at the steady state

1 1 1
dZt = dNt + dCt (1.45)
Z N C
de…ne
c dXt
xft = (1.46)
X
thus

zbt = n
bt + b
ct (1.47)
Doing the same for the production function

ybf t = n
bt + zbt (1.48)
Knowing that (without government) in equilibrium consumption equal in-
come
1.2 BASIC NEW KEYNESIAN MODEL 7

1+
ybf t = zbt (1.49)
+
Impulse response function with ‡exible prices

Sticky Prices Equlibrium

The price index and in‡ation is determined by the joint solution of the fol-
lowing dynamic equations:
Price index evolves according to:

Pt1 = (1 ) (pt )1 + Pt1 1 (1.50)

The expression for the optimal price
P1 i Pt
pt Et i=0 i;t+i M Ct+i Pt+i
= P1 1
(1.51)
Pt 1 i Pt
Et i=0 i;t+i Pt+i

Log linearising the price index, we get:

Pbt = (1 ) pbt + Pbt 1 (1.52)

Log linearising the second we get
X
1
pbt = Et i i d
M C t+i + Pbt+i (1.53)
i=0

This can be quasi-di¤erentiated to get:

pbt = d
pbt+1 + M C t + Pbt (1.54)
Combining the two equations gives:

!
Pbt Pbt+1
Pbt 1 = Pbt + M
d C t + Pbt (1.55)
1 1 1 1

Solving for in‡ation, yields:

8 CHAPTER 1 NEW KEYNESIAN MODELS

t = Et t+1
d
+ eM Ct (1.56)
where
(1 ) (1 )
e= (1.57)
Which is the New Keynesian Phillips.
Express the NKPC in term of deviation from the ‡exible price equilibrium
Recall the marginal cost is given by:
Wt
MC = (1.58)
Z t Pt
Log linearizing, we get:

d
M ct
Ct = W Pbt Zbt (1.59)
Usind the labour supply condition, ybt = n
bt + zbt
d
M ct
Ct = W Pbt (b
yt bt )
n (1.60)
1+
= ( + ) ybt zbt (1.61)
+
h i
= ( + ) ybt ybtf (1.62)

Which makes possible to write the NKPC in term of deviation from the
‡exible price equilibrium

t = Et t+1 + ybt ybtf (1.63)

where = ( + )e
Chapter 2

Optimal Monetary Policy

2.0.3 Optimal Policy Problem

X
1
1 n o
T 2 2
min Lt = Et t+ + yt+i (2.1)
y
=0
2

subject to

t = Et t+1 + kyt + "t (2.2)

yt = Et yt+1 (it Et t+1 ) + t (2.3)

Lagrangian

8 9
>
> 1 2
+ 2
yt+i + >
>
>
< 2 t+ >
=
X
1
min Lt = Et t+ [ t+ Et t+1+ kyt+ "t+ ]
i >
> >
>
=0 >
:+ >
;
t+ yt+ Et yt+1+ + (it+ Et t+1+ ) t+
(2.4)
FOC respect to it+ is equal to

t+ =0

Problem can be stated just in term of and y

9
10 CHAPTER 2 OPTIMAL MONETARY POLICY

2.0.4 Discretionary Solution

Period by period problem

1 2
min L = E t + yt2 (2.5)
y 2
subject to

t = Et t+1 + kyt + "t (2.6)

"t+1 = "t + vt ; 0< < 1; vt is iid

FOC

dL
= k t + yt = 0 (2.7)
dyt
k
yt = t (2.8)

Solution

t = Et t+1 + "t . (2.9)

+ k2 + k2
Solving forward

t = "t (2.10)
k 2 + (1 )

2.0.5 Commitment - Timeless Perspective

X
1
1 2 2
min Lt = Et t+ + yt+i (2.11)
i
=0
2
subject to

t = Et t+1 + kyt + "t

where, as before, "t+1 = "t + vt is the stochastic policy process
FOC
 11

X
1
1 2 2
min Lt = Et t+ + yt+i + t+ [ t+ Et t+1+ kyt+ "t+ ]
i
=0
2
(2.12)
FOC with respect to t and yt for =0

$
= t + t =0 (2.13)
t
$
= yt k t =0 (2.14)
yt
and for >0

$
= ( t+ + t+ t+ 1) =0 (2.15)
t+
$
= yt+ k t =0 (2.16)
yt+
FOC (2)

$
f or: 0! = yt+ k t =0 (2.17)
yt+
$
f or: = 0! = t + t =0 (2.18)
t
$
f or: > 0! = t+ + t+ t+ 1 =0 (2.19)
t+

2.0.6 Policy Inertia

Combining (2.19) and (2.17) we get

t + (yt yt 1 ) = 0
k
which can be rewritten as
k
yt = yt 1 t (2.20)
12 CHAPTER 2 OPTIMAL MONETARY POLICY

Solution
substituting in the supply equation

(yt yt 1 ) = (Et yt+1 yt ) + kyt + "t

k k
yt yt kyt = Et yt+1 yt 1 + "t
k k k k
k2 k
1+ + yt = Et yt+1 + yt 1 "t (2.21)

Second order di¤erence equation to solve with undetermined coe¢ cients

method
Undetermined Coe¢ cients Method
First posit a solution for yt that is a function of the state (yt 1; "t ) :

yt = ayt 1 + b"t (2.22)

This, togheter with the assumption of shocks following a AR(1) process, gives

Et yt+1 = ayt + b "t = a2 yt 1 + b (a + ) "t (2.23)

substituting in (2.21) we get

k2 k
1+ + (ayt 1 + b"t ) = a2 y t 1 + b (a + ) "t + yt 1 "t (2.24)

that is:

k2 k
1+ + (ayt 1 + b"t ) = 1 a2 y t 1 + b (a + ) "t (2.25)

Find the value of a and b that match the coe¢ cients:

k2
1+ + a= 1 a2 t
(2.26)

k2 k
1+ + b = b (a + ) (2.27)
 13

Equation for a quadratic - choose the root jaj < 1: The solution for b is
instead unique, i.e

k
b= (2.28)
(1 + (1 a )) + k 2
Decision rule for t is:

t = (yt yt 1 ) (2.29)
k
k
t = ayt 1 "t yt 1 (2.30)
k (1 + (1 a )) + k 2

t = (1 a) yt 1 + "t (2.31)
k (1 + (1 a )) + k 2

Under both precommitment and discretion, monetary policy completely

o¤sets the impacts of the demand shock, t , so t does not a¤ect either
yt or .

Cost shocks, "t , have di¤erent impacts under precommitment and dis-
cretion because under discretion there is a stabilization bias.

There is history dependence of optimal policy under precommitment,

but none under discretion. The history dependence is a mean by which
commitment is implemented.