Lambda Group

The Basic New Keynesian Model

F. Martin Martinez
June 24, 2023

Abstract
This document is entirely based on Gali’s book chapter 3 and featured by step-by-step solution. For further
intuition of each step look at the book.

1 Baseline
The main features of the basic new Keynesian model are:
1. Closed economy
2. There is not fiscal sector
3. There is only a production input (labor)
4. No capital accumulation
5. Perfect competition in input market (labor market)
6. Imperfect competition in goods market (differentiated goods)
7. Staggered price setting (Calvo, 1983)

2 Households
Households has to make two temporal desicions:
1. Intertemporal choice
Maximize its expected utility function (discounted) over time subject to its buget constraint.
2. Intratemporal choice
Maximize its consumption among all goods subject to a fixed expenditure.
A representative infinitely-lived household, seeking to maximize:
∞
X
V = E0 β t U (Ct , Nt )
t=0

where Ct is a consumption index given by

ϵ
ˆ 1  ϵ−1
ϵ−1
Ct = Ct (i) ϵ di (1)
0

with Ct (i) representing the quantity of good i consumed by the household in period t. Note that we assume the
existence of a continuum of goods represented by the interval [0, 1]. The period budget constraint now takes the
form

1
 ˆ 1
Pt (i)Ct (i)di + Qt Bt ≤ Bt−1 + Wt Nt + Tt f or t = 0, 1, 2, ....
0

Pt (i) : P rice of good i

ˆ 1
Pt (i)Ct (i)di = Pt Ct (2)
0

2.1 Intertemporal Choice

∞
X 1 1
M axCt ,Nt ,Bt V = E0 β t U (Ct , Nt ) , U (Ct , Nt ) = C 1−σ − N 1+φ
t=0
1−σ t 1+φ t

Pt Ct + Qt Bt ≤ Bt−1 + Wt Nt + Tt
s.t.
∞ 
1 1
X   
L = E0 βt Ct1−σ − Nt1+φ + λt (Bt−1 + Wt Nt + Tt − Pt Ct − Qt Bt )
t=0
1−σ 1+φ
FOCs:

β t Ct−σ
β t UC (Ct , Nt ) − λt Pt = β t Ct−σ − λt Pt = 0 =⇒ λt = (3)
Pt
β t Ntφ
β t UN (Ct , Nt ) + λt Wt = −β t Ntφ + λt Wt = 0 =⇒ λt = (4)
Wt
λt (−Qt ) + βλt+1 (1) = 0 =⇒ λt Qt = βλt+1 (5)
Combining eq.(3) and eq.(4) by using λ, we get:

Ct−σ Wt = Ntφ Pt Labor Supply (6)

Pluggint eq.(3) into eq.(5):

−σ
Ct−σ βEt Ct+1
Qt =
Pt Et Pt+1
" #
−σ
Ct−σ Ct+1
Qt = βEt Euler Equation (7)
Pt Pt+1

2.2 Intratemporal Choice:

The household now must decide how to allocate its consumption expenditures among
´ 1 different goods. This requires
that the consumption index Ct be maximized for any given level of expenditures 0 Pt (i)Ct (i)di.
ϵ
ˆ 1  ϵ−1
ϵ−1
M axCt (i) Ct = Ct (i) ϵ di
0
ˆ 1
s.t. Pt Ct = Pt (i)Ct (i)di
0
ˆ 1
ϵ
 ϵ−1  ˆ 1 
ϵ−1
L= Ct (i) ϵ di + λt Pt Ct − Pt (i)Ct (i)di
0 0
FOC: 1
1
ϵ−1
ˆ  ϵ−1
∂L ϵ ϵ−1 1
= Ct (i) ϵ di Ct (i)− ϵ + λt [−Pt (i)] = 0
∂Ct (i) ϵ−1 0 ϵ

2
 1 1
Ctϵ Ct (i)− ϵ = λt [Pt (i)] (8)
∂L 1 1
= 0 =⇒ Ct Ct (j)− ϵ = λt [Pt (j)] ϵ
(9)
∂Ct (j)
Dividing eq.(8) into eq.(9):
−ϵ
Pt (i)

Ct (i) = Ct (j) (10)
Pt (j)
Plugging eq.(10) into our objective function:
ϵ
ˆ # ϵ−1
 "  ϵ−1
−ϵ
Pt (i)
1 ϵ

Ct = Ct (j) di
Pt (j)

0

ϵ
Ct (j) 1
ˆ  ϵ−1
 ϵ−1
Ct = Pt (i)−ϵ ϵ di

Pt (j)−ϵ 0
ϵ
Ct (j) 1
ˆ  ϵ−1
Ct = Pt (i) 1−ϵ
di (11)
Pt (j)−ϵ 0

Plugging eq.(10) into our budget constraint:

ˆ −ϵ
1
Pt (i)

P t Ct = Pt (i) Ct (j)di
0 Pt (j)
ˆ 1
Ct (j)
Pt C t = Pt (i)1−ϵ di (12)
0 Pt (j)−ϵ
Plugging eq.(12) into eq.(11), we get:
ϵ ˆ
Ct (j) 1 1
Ct (j)
ˆ  ϵ−1
Pt Pt (i) 1−ϵ
di = Pt (i)1−ϵ di
Pt (j)−ϵ 0 0 Pt (j)−ϵ
1
ˆ 1  1−ϵ
Pt = Pt (i) 1−ϵ
di , Aggregate P rice Level (13)
0

Replacing this last equation into eq.(11):

Ct (j) −ϵ
PCt =
Pt (j)−ϵ t
−ϵ
Pt (j)

Ct (j) = Ct , ∀j ∈ [0, 1]
Pt
This equation is the same for the good i:
−ϵ
Pt (i)

Ct (i) = Ct , ∀i ∈ [0, 1] Dixit and Stiglitz ′ s demand (14)
Pt

2.3 Summary of main equations:

1. Labor supply: eq.(6)
Ct−σ Wt = Ntφ Pt

2. Euler Equation: eq.(7) " #

−σ
Ct−σ Ct+1
Qt = βEt
Pt Pt+1

3
 3. Consumption Index: eq.(1)
ϵ
ˆ 1  ϵ−1
ϵ−1
Ct = Ct (i) ϵ di
0

4. Aggregate price level: eq.(13)

1
ˆ 1  1−ϵ
Pt = Pt (i) 1−ϵ
di
0

5. Demand of goods: eq.(14)

−ϵ
Pt (i)

C(i)t = Ct
Pt

3 Firms
3.1 Production Function
We assume a coninuum of firms indexed by i ∈ [0, 1]. Each firm produces a differentiated good, but they all use an
identical technology, represented by the production function
Yt (i) = At Nt (i)1−α (15)
where At represents the level of technology, assumed to be common to all firms and to evolve exogenously over
time.
To choose the optimum level of N (i)t , we must maximize:

Ω = Pt (i)Yt (i) − Wt Nt (i)

Ω = Pt (i)At Nt (i)1−α − Wt Nt (i)
∂Ω
= (1 − α)Pt (i)At Nt (i)−α − Wt = 0
∂Nt (i)
Yt (i)
Wt = (1 − α)Pt (i)At Nt (i)−α = (1 − α)Pt (i) Labor Demand (16)
Nt (i)

3.2 Aggregate Price Dynamics

Following the formalism proposed in Calvo (1983), each firm may reset its price only with probability 1 − θ in any
given period, independently of the time elapsed since the last adjustment. Thus, each period a measure 1 − θ of
producers reset their prices, while a fraction θ keep their prices unchanged. As a result, the average duration of a
price is given by 1−θ
1
. In this context, θ becomes a natural index of sitckiness. In other words, each period, each
firm has two options:
1. Choose the optimum price with probability ”1 − θ”: P (i)t = Pt∗
2. Remain the last price with probability ”θ”: P (i)t = Pt−1
Recall eq.(13) and taking into account the probability of stickiness:
1
ˆ 1  1−ϵ
Pt = Pt (i) 1−ϵ
di
0
ˆ 1
Pt1−ϵ = Pt (i)1−ϵ di
0
ˆ θ ˆ 1
Pt1−ϵ = 1−ϵ
Pt−1 di + Pt∗1−ϵ di
0 θ
ˆ θ ˆ 1
Pt1−ϵ = Pt−1
1−ϵ
di + Pt∗1−ϵ di
0 θ

Pt1−ϵ = 1−ϵ
Pt−1 θ + Pt∗1−ϵ (1 − θ) (17)

4
3.3 Optimal Price Setting
A firm reoptimazing in period t will choose the price Pt∗ that maximizes the current market value of the profits
generated while that price remains effective.

3.3.1 Total Cost

T C(i) = Wt Nt (i) (18)
Recall eq.(15) and rearranging it in order to get an expression of Nt :
1
Yt (i)
  1−α
Nt (i) = (19)
At

Replacing this equation into total cost

1
Yt (i)
  1−α
T Ct (i) = Wt
At

• Marginal Cost:

α α
∂T Ct (i) 1 Yt (i) 1−α 1 At1−α Nt (i)α 1 Nt (i)α
M Ct (i) = = Wt = W t = Wt (20)
∂Yt (i) (1 − α) A 1−α
1
(1 − α) 1
A 1−α (1 − α) At
t t
Nt (i)
Adding Nt (i) :
1 Nt (i)α Nt (i) 1 1
M Ct (i) = Wt = Wt Nt (i)
(1 − α) At Nt (i) (1 − α) At Nt (i)1−α
1 1
M Ct (i) = Wt Nt (i)
(1 − α) Yt (i)
(1 − α)M Ct (i)Yt (i) = Wt Nt (i) (21)

Plugging eq.(21) into eq.(18) in order to get a function of total cost in terms of output units.

T Ct (i) = (1 − α)M Ct (i)Yt (i) = Ψt (Yt )

3.3.2 The Stochastic Discount Factor:

Recall the euler equation eq.(7): " #
−σ
Ct−σ Ct+1
Qt = βEt
Pt Pt+1
So, the stochastic discount factor at time t is:
" #
−σ
Ct+1 Pt
Qt = βEt
Pt+1 Ct−σ

the stochastic discount factor at time t for k periods ahead is1 :

" #
−σ
Ct+k Pt
Qt,t+k = β Et
k
(22)
Pt+k Ct−σ
h i
UC (Ct+s )
1 Stochastic discount factor is defined as:Qt+s = β s Et UC (Ct )

5
3.3.3 Firm’s Maximization Problem2
∞
( )
X
M axPt∗ Ω = Et θ Qt,t+k Pt Yt+k/t (i) − T Ct+k/t (i) (23)
k
 ∗ 

k=0

s.t. −ϵ
Pt∗

Yt+k/t (i) = Yt+k (24)
Pt+k
This constraint is based on the fact that P (i)t = Pt∗ as well as given a closed economy Yt = Ct , eq.(14) can be
rewritten in terms of Yt .
Plugging eq. (21) into eq.(23):
(∞ )
X
θ Qt,t+k Pt Yt+k (i) − (1 − α)M C(i)t+k/t Yt+k/t (i)
k
 ∗ 
Et
k=0

∞
( )
X
θ Qt,t+k Pt∗ − (1 − α)M Ct+k/t (i) Yt+k/t (i)
k
(25)
 
Et
k=0

Replacing eq.(24) into eq.(25):

∞
( −ϵ )
Pt∗
X 
Ω = Et θ Qt,t+k Pt∗ − (1 − α)M Ct+k/t (i)
k
 
Yt+k
Pt+k
k=0

∞
( " # )
X Yt+k h ∗1−ϵ ∗−ϵ
i
Ω = Et k
θ Qt,t+k −ϵ Pt − (1 − α)M Ct+k/t (i)Pt
k=0
Pt+k
FOC:
∞
( # " )
∂Ω X Yt+k h ∗−ϵ ∗−ϵ−1
i
= Et k
θ Qt,t+k −ϵ (1 − ϵ)Pt − (1 − α)M Ct+k/t (i)(−ϵ)Pt =0
∂Pt∗ Pt+k
k=0
(∞ " −ϵ
# )
∂Ω X Yt+k Pt∗ h
∗−1
i
= Et k
θ Qt,t+k −ϵ (1 − ϵ) − (1 − α)M Ct+k/t (i)(−ϵ)Pt =0
∂Pt∗ Pt+k
k=0
−ϵ
Pt∗ Yt+k Pt∗
multiplying by (1−ϵ) and taking into account that −ϵ
Pt+k
= Y (i)t+k/t

∞
(  )
∂Ω X ϵ
= Et θ Qt,t+k Yt+k/t (i) Pt∗ − (1 − α)M Ct+k/t (i)
k
=0 (26)
 
∂Pt∗ (ϵ − 1)
k=0

Notice that limiting the case of no price frictions (θ = 0) the previous condition collapses to the familiar optimal
price setting condition under flexible prices
ϵ
Pt∗ = (1 − α)M Ct+k/t (i)
(ϵ − 1)

1
 
Pt∗ = 1 − (1 − α)M Ct+k/t (i)
ϵ−1

if
ϵ > 1 =⇒ Pt∗ > (1 − α)M Ct+k/t (i)
ϵ → ∞ =⇒ Pt∗ = (1 − α)M Ct+k/t (i)
P∞  
2 The firm’s maximization problem could be rewritten acording to Gali’s notation:M axP ∗ k=0
θk Qt,t+k Pt∗ Y t+k/t (i) − Ψt+k (Yt+k/t (i))
t

6
 −ϵ−1
Dividing eq.(26) by Pt∗ , we get3 :
Dividing eq.(26) by Pt−1 :
(∞ )
 Pt∗ 1 Pt+k

∂Ω X ϵ
= Et θ Qt,t+k Yt+k/t (i)
k
− (1 − α)M Ct+k/t (i) =0

∂Pt∗ Pt−1 (ϵ − 1) Pt−1 Pt+k
k=0

M Ct+k/t (i)
Rewriting this equation: M Ct+k/t
r
(i) = (real marginal cost) and PPt+k = t−1,t+k (inf lation), we have
Q
Pt+k t−1
our optimal price setting condition4 :
(∞  ∗ )
∂Ω X P ϵ Y
= Et θk Qt,t+k Yt+k/t (i) t
− (1 − α)M Ct+k/t
r
(i) =0 (27)
 
t−1,t+k
∂Pt∗ Pt−1 (ϵ − 1)
k=0

3.4 Summary of main equations

1. Production Function: eq.(15)
Yt (i) = At Nt (i)1−α

2. Labor demand: eq.(16)

Wt = (1 − α)Pt (i)At Nt (i)−α
Yt (i)
Wt = (1 − α)Pt (i)
Nt (i)

3. Marginal Cost: eq.(20)

1 Nt (i)α
M Ct (i) = Wt
(1 − α) At

4. Aggregate price dynamics: eq.(17)

Pt1−ϵ = Pt−1
1−ϵ
θ + Pt∗1−ϵ (1 − θ)

5. The stochastic discount factor: eq.(22)

" #
−σ
Ct+k Pt
Qt,t+k = βEt
Pt+k Ct−σ

6. Optimal price stting condition: eq.(27)

(∞  ∗ )
∂Ω X P ϵ Y
= Et θk Qt,t+k Yt+k/t (i) t
− (1 − α)M Ct+k/t
r
(i) =0
 
t−1,t+k
∂Pt∗ Pt−1 (ϵ − 1)
k=0

P∞   
3 This could be rewritten acording to Gali’s notation: k=0
θk Et Qt,t+k Yt+k/t Pt∗ − ψt+k/t M = 0 ; where : M =
ϵ ′
(ϵ−1)
; ψt+k/t = Ψt+k/t (Yt+k/t ) = (1 − α)M Ct+k/t (i)
Pt∗
P∞ n  h Q io
4 Acording to Gali’s notation: k=0
θk Et Qt,t+k Yt+k/t Pt−1
− MM Ct+k/t (i) t−1,t+k
= 0 ; where : M =
ϵ ψt+k/t
(ϵ−1)
; M Ct+k/t (i) = Pt+k

7
4 Log Linear Techniques
Handling and solving models with substantial nonlinearity is often difficult. Linear models are often much easier
to solve, and there exist well-developed methods for solving linear models. The problem is to convert a nonlinear
model into a sufficiently good linear approximation so that the solutions to the linear approximation are helpful in
understanding the behavior of the underlying nonlinear system. A now standard method for linear approximation
is to log-linearize a model around its stationary state. The assumption is that, if the model is not too far from the
stationary state, the linear version that results closely approximates the original model.

4.1 The Basic of Log Linearization

Consider a nonlinear model:
G(xt )
F (xt ) =
H(xt )
where xt is a vector of the variables of the model that can include expectational variables and lagged variables
in addition to contemporaneous variables. The process of log linearization is to first take the logarithms of the
function F (), G(), and H(), and then take a first-order Taylor approximation. Taking the logarithms gives
ln(F (x)) = ln(G(x)) − ln(H(x)) (28)
Taylor approximation:
∞
X ∂ n F (x) (x − x)n
F (x) = | x=x
n=0
∂xn n!
First-order Taylor approximation:
∂F (x)
F (x) ∼
= F (x)|x=x + |x=x (x − x)
∂x
Taking the first-order Taylor approximation of eq. (28) around the steationary state values , x, gives
F ′ (x) G′ (x) H ′ (x)
ln(F (x)) + (xt − x) ≡ ln(G(x)) + (xt − x) − ln(H(x)) − (xt − x)
F (x) G(x) H(x)
Given that the log version of the model holds at the stationary state,
ln(F (x)) = ln(G(x)) − ln(H(x))
we can eliminate the three ln() components, and the equation simplifies to:
F ′ (x) G′ (x) H ′ (x)
(xt − x) ≡ (xt − x) − (xt − x)
F (x) G(x) H(x)
This implicit assumption is that one is staying close enough to the stationary state, x, so that the second-order
or higher terms of the Taylor expansion are small enough to be irrelevant and can safely be left out.

4.2 Uhlig’s Method of Log Linearization

Harald Uhlig’s recommends using a simpler method for finding log-linear approximations of functions. His method
does not require taking derivates and gives the same result as the above method, except that the linear model is
expressed in terms of log differences of the variables.
Consider and equation of a set of variables Xt . Define xt = lnXt − lnX. The lowercase variables are the lod
difference of the original variables from the value X. One can write the original variable as
Xt = Xext = X(1 + xt )

Xt − X
xt = lnXt − lnX ∼
=
X
since
Xt
Xt


ln
Xext = XelnXt −lnX = Xe X =X = Xt
X

8
5 Log Linear Model
In this section, we will log linearize some main equations both for households and firms:

5.1 Households
1. Labor supply: eq.(6)
Ct−σ Wt = Ntφ Pt =⇒ wt − pt = φnt + σct (29)
2. Euler Equation: eq.(7) " # " #
−σ −σ
Ct−σ Ct+1 −σ Ct+1
Qt = βEt =⇒ Ct Qt = βEt Q (30)
Pt Pt+1 t,t+1

Taking into account that Qt = 1+i 1

t
and (1 + it ) = eit , we can conclude that Qt = e−it as well as that inflation
in the steady state is equal to 1.
1
=⇒ ct = Et ct+1 − [it − Et πt,t+1 − ρ] where ρ = −logβ (31)
σ
5.2 Firms
1. Production Function: eq.(15)
Yt (i) = At Nt (i)1−α =⇒ yt (i) = at + (1 − α)nt (i) (32)

2. Labor demand: eq.(16)

Wt = (1 − α)P (i)t At Nt (i)−α =⇒ wt − pt = at − αnt (i) + log(1 − α) (33)
Yt (i)
Wt = (1 − α)Pt (i) =⇒ wt − pt = yt (i) − nt (i) + log(1 − α) (34)
Nt (i)
3. Marginal Cost: eq.(20)
1 Nt (i)α
M Ct (i) = Wt
(1 − α) At
in real terms
M Ct (i) 1 Nt (i)α Wt
M Ctr (i) = =
Pt (1 − α) At Pt

=⇒ mcrt (i) = αnt (i) + wt − at − pt − log(1 − α) (35)

4. Aggregate price dynamics: eq.(17)
Pt1−ϵ = Pt−1
1−ϵ
θ + Pt∗1−ϵ (1 − θ)

1−ϵ 1−ϵ 1−ϵ

P (1 + (1 − ϵ)pt ) = P (1 + (1 − ϵ)pt−1 )θ + P (1 + (1 − ϵ)p∗t )(1 − θ)

at steady state:
1−ϵ 1−ϵ 1−ϵ
P =P θ+P (1 − θ)
so
1−ϵ 1−ϵ 1−ϵ
P (1 − ϵ)pt = P (1 − ϵ)pt−1 θ + P (1 − ϵ)p∗t (1 − θ)
=⇒ pt = pt−1 θ + p∗t (1 − θ) (36)
substracting pt−1 in both sides
pt − pt−1 = pt−1 θ − pt−1 + p∗t (1 − θ)
=⇒ πt = (1 − θ)(p∗t − pt−1 ) where πt = pt − pt−1 (37)
both eq.(36) and eq.(37)5 are the same.
5 Equation (7) in Gali’s book.

9
5. The stochastic discount factor: eq.(22)
" #
−σ
Ct+k Pt
Qt,t+k = β Et k
Pt+k Ct−σ

at steady state (Ct+k = Ct ; Pt+k = Pt )

Qt,t+k = β k (38)

so
qt,t+k = −σ (Et ct+k − ct ) − (Et pt+k − pt ) + klogβ

=⇒ qt,t+k = −σ (Et ct+k − ct ) − (Et πt,t+k ) + klogβ (39)

6. Optimal price setting condition: eq.(27)

Pt∗
P∞ k h oi
∂Pt∗ = Yt+k/t (i) (i) (ϵ−1) =0
∂Ω r ϵ
Q
k=0 θ Et {Qt,t+k Pt−1 −(1 − α)M Ct+k/t t−1,t+k
̸= 0 ̸= 0 =0
Pt∗ ϵ Y
= (1 − α)M Ct+k/t
r
(i) t−1,t+k
Pt−1 (ϵ − 1)

This equation in the steady state: Qt,t+k = β k (eq.(22) =⇒ Ct+k = Ct and Pt+k = Pt = Pt−1 = Pt∗ ) and in turn =
Q
1
r ϵ−1
M C t+k/t (i) = (40)
ϵ (1 − α)
Now, in log linear terms must be (using Uhlig’s Method):
∞  i  P ep∗t 
X h r mcrt+k/t (i) ϵ P pt+k −pt−1
θk Et Qt,t+k eqt,t+k Y eyt+k/t (i) − (1 − α)M C (i)e e =0
P ept−1
t+k/t
(ϵ − 1) P
k=0

Taking into account eq.(38) and eq.(40):

∞
ϵ−1
  
X k p∗ ϵ mcrt+k/t (i)+pt+k −pt−1
(θβ) Y Et e qt,t+k +yt+k/t (i)
e t −pt−1
− (1 − α) e =0
ϵ (1 − α) (ϵ − 1)
k=0

∞ nh io
X ∗ r
eqt,t+k +yt+k/t (i)+pt −pt−1 − eqt,t+k +yt+k/t (i)+mct+k/t (i)+pt+k −pt−1
k
Y (θβ) Et =0
k=0
∞
X nh io
k
(θβ) Et (1 + qt,t+k + yt+k/t (i) + p∗t − pt−1 ) − (1 + qt,t+k + yt+k/t (i) + mcrt+k/t (i) + pt+k − pt−1 ) = 0
k=0
∞
X nh io
k
(θβ) Et p∗t − pt−1 − (mcrt+k/t (i) + pt+k − pt−1 ) =0
k=0
∞
X ∞
X n o
k k
(θβ) Et {p∗t − pt−1 } = (θβ) Et mcrt+k/t (i) + pt+k − pt−1 (41)
k=0 k=0
∞
X ∞
X n o
k k
(p∗t − pt−1 ) (θβ) = (θβ) Et mcrt+k/t (i) + pt+k − pt−1
k=0 k=0
∞
1 X k
n o
(p∗t − pt−1 ) = (θβ) Et mcrt+k/t (i) + pt+k − pt−1
1 − θβ
k=0
∞
X n o
k
=⇒ (p∗t − pt−1 ) = (1 − θβ) (θβ) Et mcrt+k/t (i) + pt+k − pt−1 (42)
k=0

10
5.3 Equlibrium without Frictions (log-linear version)
1. Market Clearing: the superscript n implies an equilibrium without frictions.

ytn (i) = cnt (i)

the superscript n implies an equilibrium without frictions.

2. Labor Market (eq.(29)=eq.(34))
Labor Demand = Labor Supply
wt − pt = ytn (i) − nt (i) wt − pt = φnt (i) + σcnt (i)

ytn (i) − nt (i) = φnt (i) + σcnt (i)

taking into account market clearing condition

ytn (i) − nt (i) = φnt (i) + σytn (i)

1−σ n
nt (i) = y (i) (43)
1+φ t

3. Production Function eq.(32)

ytn (i) = at + (1 − α)nt (i)

replacing eq.(43) into log-linear production function

1−σ n
 
ytn (i) = at + (1 − α) y (i)
1+φ t

(1 + φ)at
ytn (i) = (44)
φ + α + σ(1 − α)

6 The Phillips Curve

In this section, we will obtain one of the key building blocks of the new Keynesian model. The new Keynesian
Phillips curve is the one which relates inflation to its one period ahead forecast and the output gap.
1. First of all, recall the log-linear real marginal cost eq.(35)

mc(i)rt = αnt (i) + wt − at − pt − log(1 − α)

this eqaution k periods ahead is

mcrt+k (i) = αnt+k (i) + wt+k − at+k − pt+k − log(1 − α) (45)

eq.(45) is the marginal cost in the period t + k. However, the marginal cost that firms take into account at
the moment of deciding its optimum price level (Pt∗ ) is the one that they can calculate at time t, so

mcrt+k/t (i) = αnt+k/t (i) + wt+k − at+k − pt+k − log(1 − α) (46)

Substracting eq.(46) to eq.(45)

mcrt+k/t (i) − mcrt+k (i) = α nt+k/t (i) − nt+k (i) (47)

11
 • As we know, we can rewrite nt in terms of yt using the production function:

Yt (i) = At Nt (i)1−α

recall its log-linear version

yt (i) = at + (1 − α)nt (i)
yt (i) − at
nt (i) = (48)
(1 − α)
this equation k periods ahead
yt+k (i) − at+k
nt+k (i) =
(1 − α)
taking into account that if nt+k is calculated in period t + k, this implies that eq.(24) is equivalent to
Yt+k (i) = Yt+k since Pt ∗ = Pt+k , So
yt+k − at+k
nt+k (i) = (49)
(1 − α)
now, we will obtain the level of nt+k calculated in the time t
yt+k/t (i) − at+k
nt+k/t (i) = (50)
(1 − α)

• Likewise, recall eq.(24)

−ϵ
Pt∗

Yt+k/t (i) = Yt+k
Pt+k
this equation in log-linear terms is

yt+k/t (i) − yt+k = −ϵ [p∗t − pt+k ] (51)

plugging eq.(49) and eq.(50) into eq.(47):

yt+k/t (i) − at+k
 
yt+k − at+k
mcrt+k/t (i) − mcrt+k (i) =α −
(1 − α) (1 − α)
α
mcrt+k/t (i) − mcrt+k (i) = yt+k/t (i) − yt+k (52)


1−α
plugging eq.(51) into eq.(52), we get6 :
α
mcrt+k/t (i) − mcrt+k (i) = (−ϵ [p∗t − pt+k ])
1−α
αϵ
=⇒ mcrt+k/t (i) = mcrt+k (i) − (p∗ − pt+k ) (53)
1−α t
2. Now, we are going to reduce eq.(41) by dropping −pt−1 in both sides and using eq.(53)
∞
X ∞
X n o
k k
(θβ) Et {p∗t − pt−1 } = (θβ) Et mcrt+k/t (i) + pt+k − pt−1
k=0 k=0

∞ ∞  
X k
X k αϵ
(θβ) Et {p∗t } = (θβ) Et mcrt+k (i) − (p∗ − pt+k ) + pt+k
1−α t
k=0 k=0

rearranging terms
∞    ∞   
X k αϵ X k αϵ
(θβ) Et p∗t 1+ = (θβ) Et mcrt+k (i) + pt+k 1+
1−α 1−α
k=0 k=0
6 This equation in Gali’s book is the equation (14)

12
 ∞ ∞
1 − α + αϵ 1 − α + αϵ
X    X   
k k
(θβ) Et p∗t = (θβ) Et mcrt+k (i) + pt+k
1−α 1−α
k=0 k=0

multiplying both sides by 1−α

1−α+αϵ

∞ ∞
1−α X 1 − α + αϵ 1 − α + αϵ 1−α
   X   
k k
(θβ) Et pt
∗
= (θβ) Et mct+k (i) + pt+k
r
1 − α + αϵ 1−α 1−α 1 − α + αϵ
k=0 k=0

∞ ∞
1−α
X X  
k k
(θβ) Et {p∗t } = (θβ) Et mcr (i) + pt+k
1 − α + αϵ t+k
k=0 k=0

substracting pt−1 in both sides

∞ ∞
1−α
X X  
k k
(θβ) Et {p∗t − pt−1 } = (θβ) Et mcr (i) + pt+k − pt−1
1 − α + αϵ t+k
k=0 k=0

∞
1 1−α
X  
k
(p∗t − pt−1 ) = (θβ) Et mcr (i) + pt+k − pt−1
1 − θβ 1 − α + αϵ t+k
k=0
∞
X k 1−α
p∗t − pt−1 = (1 − θβ) (θβ) Et Θmcrt+k (i) + pt+k − pt−1 where Θ = <1 (54)

1 − α + αϵ
k=0

3. In this part, we will obtain an equation which relates inflation to its one period ahead forecast and the output
gap.

(a) Dropping pt−1 in eq.(54) and expanding it

∞
X k
p∗t = (1 − θβ) [Θmcrt (i) + pt ] + (1 − θβ) (θβ) Et Θmcrt+k (i) + pt+k (55)


k=1

(b) Dropping pt−1 in eq.(54) and using it in a period ahead

∞
X k
Et p∗t+1 = (1 − θβ) (θβ) Et Θmcrt+k+1 (i) + pt+k+1


k=0

this could be rewritten as

∞
1 X k
Et p∗t+1 = (1 − θβ) (θβ) Et Θmcrt+k (i) + pt+k

θβ
k=1

∞
X k
Et p∗t+1 θβ = (1 − θβ) (θβ) Et Θmcrt+k (i) + pt+k (56)


k=1

(c) Replacing eq.(56) into eq.(55)

p∗t = (1 − θβ) [Θmcrt (i) + pt ] + Et p∗t+1 θβ (57)

substracting pt−1 in both sides

p∗t − pt−1 = (1 − θβ) [Θmcrt (i) + pt ] + Et p∗t+1 θβ − pt−1

p∗t − pt−1 = Θmcrt (i) + pt − θβΘmcrt (i) − θβpt + Et p∗t+1 θβ − pt−1

p∗t − pt−1 = [1 − θβ] Θmcrt (i) + θβ Et p∗t+1 − pt + pt − pt−1
 

p∗t − pt−1 = [1 − θβ] Θmcrt (i) + θβ Et p∗t+1 − pt + πt

 

13
 multipliying by (1 − θ) in both sides

(1 − θ) [p∗t − pt−1 ] = [1 − θβ] (1 − θ)Θmcrt (i) + θβ(1 − θ) Et p∗t+1 − pt + (1 − θ)πt

 

recall eq.(37):πt = (1 − θ)(p∗t − pt−1 )

πt = [1 − θβ] (1 − θ)Θmcrt (i) + θβEt πt+1 + (1 − θ)πt

[1 − θβ] (1 − θ)
πt = Θmcrt (i) + βEt πt+1
θ
so7
[1 − θβ] (1 − θ)
πt = λmcrt (i) + βEt πt+1 ; where λ = Θ (58)
θ
(d) Now, we are going to find an equation of mcrt (i) in terms of output gap (e
yt = yt (i) − ytn (i))
recall eq.(29) and eq.(35)
wt − pt = φnt (i) + σct (i)

mcrt (i) = αnt (i) + wt − at − pt − log(1 − α)

combining both equations (assuming equlibrium in labor market and market clearing yt (i) = ct (i))

mcrt (i) = αnt (i) − at + φnt + σyt (i)

mcrt (i) = (α + φ) nt (i) − at + σyt (i)

now, we recall that the level of nt (i) in terms of yt (i) is (eq.(48) )

yt (i) − at
nt (i) =
(1 − α)
replacing terms
yt (i) − at
 
mcrt (i)= (α + φ) − at + σyt (i)
(1 − α)
yt (i)
   
at
mct (i) = (α + φ)
r
− (α + φ) − at + σyt (i)
(1 − α) (1 − α)
σ(1 − α) + α + φ 1+φ
   
mcrt (i) = yt (i) − at
1−α (1 − α)
σ(1 − α) + α + φ 1+φ
  
mct (i) =
r
yt (i) − at
1−α σ(1 − α) + α + φ
(1+φ)at
recall eq.(44) :ytn (i) = φ+α+σ(1−α) , so

σ(1 − α) + α + φ
 
mcrt (i)
= [yt (i) − ytn (i)]
1−α

σ(1 − α) + α + φ
 
mc(i)t =
r
yet ; where yet = yt (i) − ytn (i) (59)
1−α
(e) Finally, replacing eq.(59) into eq.(58), we get8

πt = λmc(i)rt + βEt πt+1

σ(1 − α) + α + φ
 
πt = λ yet + βEt πt+1
1−α
7 Equation (16) in Gali’s book.
8 Equation (21) in Gali’s book.

14
 πt = κe
yt + βEt πt+1 (60)
where
σ(1 − α) + α + φ
 
κ=λ
1−α
[1 − θβ] (1 − θ)
λ= Θ
θ
1−α
Θ=
1 − α + αϵ
Equation (60) is the new Keynesian Phillips curve.

7 The IS Curve
In this section, we will obtain the second key equation describing the equilibrium of the New Keynesian model.
As mentioned earlier, this basic framework of NEK model assumes a closed economy:
Yt = Ct
and in its log-linear version
yt = ct
Using this consition into the log-linear euler equation eq.(31) (in its natural level-without frictions)
1n
ytn = Et yt+1 (61)
 n n

− i − Et πt,t+1 −ρ
σ t
Using Fisher equation:
rtn = int − Et πt+1
n
(62)
Plugging eq.(61) into and rearranging terms
−σ ytn − Et yt+1 = it − Et πt,t+1
  n  n n

−ρ
− yt = [rtn − ρ]
  n 
σ Et yt+1
n
σEt △yt+1 + ρ = rtn (63)
given that the economy is not in its natural level, Fisher equation and Euler equation in term of output are
1
yt = Et {yt+1 } − [it − Et πt,t+1 − ρ] (64)
σ
rt = it − Et πt+1 (65)
adding Et yt+1 − yt both sides into eq.(61) and rearranging terms:
n n

1
yt + Et yt+1
n
− ytn = Et {yt+1 } − [it − Et πt,t+1 − ρ] + Et yt+1
n
− ytn



σ
1
yt − ytn = Et yt+1 − yt+1
n
− [it − Et πt,t+1 − ρ] + Et yt+1
n
− ytn (66)

σ
defining the output gap as follow
yet = yt − ytn
so, eq.(66) is
1
yet = Et {e [it − πt,t+1 ] + Et yt+1
n
− ytn
 
yt+1 } −
σ
1
yet = Et {e
yt+1 } − [it − πt,t+1 ] + Et △yt+1 n
 
σ
using eq.(63) and eq.(65)
1 1
yet = Et {e
yt+1 } − [it − πt,t+1 ] + rtn
σ σ
Finally, we get our dynamic IS equation
1
yet = Et {e
yt+1 } − [it − πt,t+1 − rtn ] (67)
σ

15
8 A Monetary Policy Rule
To analyze an equilibrium, we should set a simple interest rate rule:

it = ρ + ϕπ πt + ϕy yet + υt

where υt is an exogenous (possibly stochastic) component with mean zero. We assume ϕπ and ϕy are non-negative
coefficients, chosen by the monetary authority. Note that the choice of the intercept ρmakes the rule consistent
with zero inflation steady state.

9 Three main equations

The Dynamic IS
1
yet = Et {yet+1 } − [it − πt,t+1 − rtn ]
σ
The New Keynesian Phillips Curve
πt = κyet + βEt πt+1

A Monetary Policy Rule

it = ρ + ϕπ πt + ϕy yet + υt

16