The New Keynesian Model

Valerio Nispi Landi∗

First version: November 2018

This version: March 2021

1 The Model1
The New Keynesian model is an extension of the RBC model with two main differences. First,
there is a continuum of intermediate-good firms that produce a differentiated intermediate
good and act under monopolistic competition. The differentiated input is sold to final-good
firms, which operate under perfect competition. Final-good firms use the differentiated goods
as inputs to produce a final good, which is sold to households. Second, intermediate-good firms
pay adjustment costs when they change prices. This nominal rigidity breaks down the classical
dichotomy: unlike the RBC model, real and nominal variables cannot be analyzed separately
in the NK model and monetary policy can affect real variables. I use the same notation of the
lecture notes on the RBC model, new variables are defined when they first show up.2

1.1 Households
The representative household solves the following optimization problem:
∞
ct1−σ h1+ϕ
X  
max E0 β t
− κL t
{ct ,it ,ht ,kt ,bt }∞
t=0
t=0
1−σ 1+ϕ

 pt ct + pt it + bt = rtk pt kt−1 + rt−1 bt−1 + pt wt ht − pt tt + pt Γt

s.t.
  2 
κ i
kt = (1 − δ) kt−1 + 1 − 2I it−1
 t
−1 it

∗
Bank of Italy, International Relations and Economics Directorate. Email: [Link]@[Link]
1
In these lecture notes I derive the equations used to simulate the model in the Dynare file [Link] and I
explain how to compute the steady state. The notes are preliminary, if you find any error or inaccuracies, feel
free to contact me. The views expressed in these notes are those of the author and do not necessarily reflect
those of the Bank of Italy.
2
This model is a simplified version of Smets and Wouters (2003) and Christiano et al. (2005). In particular,
compared to Smets and Wouters (2003), this model does not feature habits in consumption, wage rigidity, and
capital utilization. For a textbook treatment, the interested reader can refer to Woodford (2003) and Gali
(2015).

1
where pt is the price level, bt is the amount of a one-period nominal risk-free bond paying
a nominal gross interest rate rt . Notice that the budget constraint is expressed in nominal
terms. Divide both the left and right-hand side of the budget constraint by pt and write the
Lagrangian function as follows:
(∞
c1−σ h1+ϕ
   
X t bt bt−1
L = E0 β t
− κL t − λt ct + it + − rtk kt−1 − rt−1 − wt ht + tt − Γt
t=0
1−σ 1+ϕ pt pt
( "  # )))
 2
κI it
−qt λt kt − (1 − δ) kt−1 − 1 − −1 it .
2 it−1

First order conditions (foc) with respect to (wrt) consumption:

c−σ
t = λt . (1)

Foc wrt labor:

κL hϕt = λt wt . (2)

Foc wrt bonds:  

rt
λt = βEt λt+1 , (3)
πt+1
pt
where πt ≡ pt−1
is the gross inflation rate. Foc wrt capital:

k
 
0 = −qt λt + βEt λt+1 rt+1 + qt+1 λt+1 (1 − δ)
 
λt+1  k 
qt = βEt rt+1 + (1 − δ) qt+1 . (4)
λt

Foc wrt investment:

"     2 #
it it κI it
0 = −λt + qt λt 1 − κI −1 − −1 +
it−1 it−1 2 it−1
( "  2  #)
it+1 it+1
+βEt qt+1 λt+1 κI −1
it it
"  2   # ( " 2  #)
κI it it it λt+1 it+1 it+1
1 = qt 1− − 1 − κI −1 + κI βEt qt+1 −1 .
2 it−1 it−1 it−1 λt it it

(5)

The real interest rate rtr is defined as follows:

rt
rtr ≡ . (6)
Et [πt+1 ]

2
1.2 Final-good firms3
The representative final-good firm uses the following CES aggregator to produce yt :
ε
Z 1 ε−1
 ε−1
yt = yt (i) ε di
0

where yt (i) is an intermediate input produced by the intermediate firm i, whose price is pt (i).
The problem of the final-good firm is the following:
Z 1
max pt y t − pt (i) yt (i) di
yt ,{yt (i)}i∈[0,1] 0
ε
Z 1 ε−1
 ε−1
s.t yt = yt (i) ε di
0

Plug the constraint in the objective function:

ε
Z 1 ε−1
 ε−1 Z 1
max pt yt (i) ε di − pt (i) yt (i) di.
{yt (i)}i∈[0,1] 0 0

Foc wrt the generic input i:

1
Z 1 ε−1
 ε−1
1
pt yt (i) ε di yt (i)− ε = pt (i)
0
1 1
pt ytε yt (i)− ε = pt (i)
 −ε
pt (i)
yt (i) = yt . (7)
pt

Now we derive an expression for the price level as a function of the price of the intermediate
goods. The price level is defined as the price of one unit of the final-good. Hence:
Z 1 
pt = min pt (i) yt (i) di s.t. yt = 1.
{yt (i)}i∈[0,1] 0

The Lagrangian reads:

(Z ε )
Z 1 1 ε−1
 ε−1
LP = pt (i) yt (i) di − ζ yt (i) ε di −1 ,
0 0

where ζ is the lagrangian multiplier. Foc wrt the generic input i:

1
Z 1 ε−1
 ε−1
− 1ε
pt (i) = ζyt (i) yt (i) ε di .
0

3
One can also assume that there is only a final-good sector, characterized by monopolistic competition and
price rigidity. In this case, the choice between differentiated goods is up to households, as in Gali (2015).

3
Rearrange:
ε
Z 1 ε−1
 −1
ε ε −1
pt (i) = ζ yt (i) yt (i) ε di
0

pt (i)ε = ζ ε yt (i)−1

yt (i) = ζ ε pt (i)−ε . (8)

Use the constraint to find an expression for ζ:

ε
Z 1 ε−1
 ε−1
yt (i) ε di = 1
0
ε
Z 1  ε−1
ζ ε−1 pt (i)1−ε di = 1
0
1
Z 1  1−ε
1−ε
ζ = pt (i) di . (9)
0

Plug (9) and (8) in the objective function:

Z 1 
pt = pt (i) yt (i) di
0
Z 1 
ε −ε
pt = pt (i) ζ pt (i) di
0
Z 1 
ε 1−ε
pt = ζ pt (i) di
0
ε Z
Z 1  1−ε 1 
1−ε 1−ε
pt = pt (i) di pt (i) di
0 0
1
Z 1  1−ε
1−ε
pt = pt (i) di , (10)
0

which is an expression for the price level as a function of the price of intermediate goods. Notice
that by using (7) and (10) one can show that real profits ΓFt for the final-good firm are zero in

4
equilibrium:
 Z 1 
1
ΓFt = pt yt − pt (i) yt (i) di
pt 0
" Z 1  −ε #
1 pt (i)
= pt y t − pt (i) yt di
pt 0 pt
 Z 1 
1 ε 1−ε
= pt yt − pt yt pt (i) di
pt 0

1
pt yt − pεt yt pt1−ε


=
pt
1
= (pt yt − pt yt )
pt
= 0.

1.3 Intermediate-good firms

There is a continuum of firms of measure unity, indexed by i, producing a differentiated input
through the following Cobb-Douglas function:

yt (i) = at (kt−1 (i))α (ht (i))1−α , (11)

where at is the total factor productivity, which follows an autoregressive process:

log (at ) = (1 − ρa ) log (a) + ρa log (at−1 ) + vta (12)

and vta ∼ N (0, σa2 ) is a technology shock. Firms operate in monopolistic competition, so they
set the price of their own good subject to the demand of the final good firm (7). In addition,
firms pay quadratic adjustment costs ACt (i) in nominal terms à la Rotemberg (1982), whenever
they adjust prices with respect to the inflation target π:4
 2
κP pt (i)
ACt (i) = −π pt y t .
2 pt−1 (i)

4
The pricing framework à la Calvo (1983) is another way to introduce a nominal rigidity, whereby only
a fraction of firms are allowed to reset the price in every period. The Calvo framework is often found in the
optimal monetary policy literature. In these lecture notes I prefer to use the Rotemberg framework, which
is easier to implement in Dynare. Under some conditions, up to a linear approximation the two frameworks
yield identical expressions. For a thorough analysis of the differences between the Calvo and the Rotemberg
framework, the reader can refer to Ascari and Rossi (2012).

5
The profit maximization problem of the generic firm i, expressed in terms of the domestic price
index, is the following:
(∞ "  2 #)
X λt p t (i) κP p t (i)
max E0 βt yt (i) − wt ht (i) − rtk kt−1 (i) − − π yt
{pt (i),ht (i),kt−1 (i),yt (i)}∞
t=0
t=0
λ0 pt 2 pt−1 (i)
  −ε
yt (i) = yt pt (i)
pt
s.t.
y (i) = a (k (i))α (h (i))1−α .
t t t−1 t

Eliminate one constraint and write the lagrangian as follows:

(∞ " 1−ε  2
I
X λt pt (i) κP pt (i)
L = E0 βt yt − wt ht (i) − (i) − rtk kt−1 −π yt +
t=0
λ0 pt 2 pt−1 (i)
"  −ε ##)
p t (i)
+mct (i) at (kt−1 (i))α (ht (i))1−α − yt ,
pt

where mct (i) is the lagrangian multiplier, which can be interpreted as the real marginal cost
of producing an additional unit of output. Foc wrt capital:

rtk = mct (i) αat (kt−1 (i))α−1 (ht (i))1−α .

Foc wrt to labor:

wt = mct (i) (1 − α) at (kt−1 (i))α (ht (i))−α .

Foc wrt to pt (i):

 −ε    −ε−1
pt (i) yt κP pt (i) yt pt (i)
(1 − ε) − − π yt + εmct (i) +
pt pt pt−1 (i) pt−1 (i) pt pt
"  2 #
λt+1 pt+1 (i) pt+1 (i)
+βEt κP − π yt+1 = 0.
λt pt (i)2 pt (i)

In a symmetric equilibrium, firms choose the same price, same inputs and same output. Using
the production function it turns out:

yt
rtk =mct α (13)
kt−1
yt
wt =mct (1 − α) . (14)
ht

6
Rearrange the pricing condition:
     
yt κP pt yt λt+1 pt+1 pt+1
(1 − ε) − − π yt + εmct + βEt κP 2 − π yt+1 = 0
pt pt−1 pt−1 pt λt pt pt
     
pt pt λt+1 pt+1 pt+1 yt+1
(1 − ε) − κP − π + εmct + βEt κP −π = 0
pt−1 pt−1 λt pt pt yt
     
pt pt λt+1 pt+1 pt+1 yt+1
(ε − 1) + κP − π − εmct − βEt κP −π = 0
pt−1 pt−1 λt pt pt yt

     
pt pt λt+1 pt+1 yt+1
pt+1 1
−π = βEt −π + (1 − ε + εmct )
pt−1 pt−1 λt pt pt yt κP
   
λt+1 yt+1 ε ε−1
πt (πt − π) = βEt πt+1 (πt+1 − π) + mct − , (15)
λt yt κP ε

which is the non-linear Phillips curve. Using (13) and (14) one can derive real profits for
intermediate firms in a symmetric equilibrium:
 2
pt (i) κP pt (i)
ΓIt = k
yt (i) − wt ht (i) − rt kt−1 (i) − − π yt
pt 2 pt−1 (i)
κP
ΓIt = yt − wt ht − rtk kt−1 − (πt − π)2 yt
2
κP
ΓIt = yt − mct (1 − α) yt − mct αyt − (πt − π)2 yt
2
h κ P
i
ΓIt = yt 1 − mct − (πt − π)2 .
2

1.4 Policy
The government finances public expenditure gt by raising lump-sum taxes:

gt = tt ,

where gt follows an autoregressive process:

log (gt ) = (1 − ρg ) log (g) + ρg log (gt−1 ) + vtg (16)

and vtg ∼ N 0, σg2 is a public spending shock. Moreover, the monetary authority sets the

nominal interest rate according to the following Taylor rule:

"  φ #1−ρr
rt  rt−1 ρr  πt φπ yt y
= exp (vtm ) , (17)
r r π y

where vtm ∼ N (0, σm

2
) is a monetary policy shock.

7
1.5 Market clearing
Clearing in the good market implies:

κP
yt = ct + it + gt + (πt − π)2 yt . (18)
2

Clearing in the bond market implies:

bt = 0.

By Walras Law, if n−1 markets are in equilibrium, the nth market is in equilibrium too. Hence,
if we impose equilibrium in every market of the economy (i.e. bond and good), one equation is
redundant. We can check if this is true by rearranging the budget constraint with the equations
derived above and the expression for total real profits Γt = ΓFt + ΓIt :

bt bt−1
ct + i t + = rtk kt−1 + rt−1 + wt ht − tt + Γt
pt pt
ct + it = mct (1 − α) yt + mct αyt − gt + ΓFt + ΓIt
h κP 2
i
ct + it = mct (1 − α) yt + mct αyt − gt + yt 1 − mct − (πt − π) yt
2
κP
ct + it + gt + (πt − π)2 yt = yt ,
2

which is the good market clearing condition.

2 Equilibrium
The equilibrium conditions of the model are the following:

λt = c−σ
t
 
λt+1 rt
1 = βEt
λt πt+1
(  k )
λt+1 rt+1 + (1 − δ) qt+1
1 = βEt
λt qt

κL hϕt = λt wt
"  2 #
κI it
kt = (1 − δ) kt−1 + 1 − −1 it
2 it−1
"  2   #
κI it it it
1 = qt 1 − − 1 − κI −1 +
2 it−1 it−1 it−1
( " 2  #)
λt+1 it+1 it+1
+κI βEt qt+1 −1
λt it it

8
 rt
rtr =
Et [πt+1 ]
α
yt = at kt−1 h1−α
t

(1 − α) mct yt = wt ht

αmct yt = rtk kt−1

   
λt+1 yt+1 ε ε−1
πt (πt − π) = βEt πt+1 (πt+1 − π) + mct −
λt yt κP ε
κP
yt = ct + it + gt + (πt − π)2 yt
2
 r ρr  π φπ  y φy 1−ρr
" #
rt t−1 t t
= exp (vtm )
r r π y

log (at ) = (1 − ρa ) log (a) + ρa log (at−1 ) + vta

log (gt ) = (1 − ρg ) log (g) + ρg log (gt−1 ) + vtg .

There are 15 equations for 15 endogenous variables:

Xt ≡ λt , ct , rtr , rtk , wt , ht , yt , kt , qt , it , rt , mct , πt , gt , at .

 

The model features 3 exogenous shocks: vt ≡ [vta , vtg , vtm ] . Compared to the RBC model de-
scribed in the lecture notes, there are 3 additional equations: the definition of the real interest
rate (6), the Phillips curve (15) and the Taylor rule (17). The 3 additional variables are rt ,
mct and πt . Notice that the price level pt (together with the definition of inflation πt ≡ ptp−1
t
) is
not included in the set of equilibrium conditions because it would introduce a unit root in the
model.

3 Steady State
Variables without time index denote the steady state level. Equation (12) and (16) in steady
state imply:

a = a

g = g.

Parameter a just affects the scale of the economy: I will calibrate it in order to normalize y = 1.
Moreover, I set h = 13 and I will compute κL ex post. In steady state the (17) implies:

π = π.

9
So by (3) one can find the steady-state nominal rate:

π
r=
β

and by using (6) the real rate reads:

1
rr = .
β
In steady state (5) implies:
q = 1,

which gives according to (4):

1
rk = − (1 − δ) .
β
By (15), in the steady state marginal costs are the following:

ε−1
mc = .
ε

Once we have rk and mc, we can get the steady state of k by (13):

αy ε − 1
k= ,
rk ε

and in turn we get i from the law of motion of capital:

i = δk.

Using (14) we can find w:

(1 − α) y ε − 1
w= .
h ε
Using (18), the steady-state level of consumption is:

c = y − i − g.

Marginal utility of consumption:

λ = c−σ .

Using (2) one can recover the value for κL :

w
κL = .
cσ hϕ

Finally, using (11), we can get the value for the calibration of a consistent with y = 1:

y
a= .
k α h1−α

10
Notice that adjustment costs are zero in the steady state, hence in the steady state the unique
distortion is the presence of monopolistic competition. Eliminating monopolistic competition,
I could get:
ε → ∞ , mct → 1,

and I could obtain the same steady state of the RBC model.5

Bibliography
Ascari, G. and Rossi, L. (2012). Trend Inflation and Firms Price-Setting: Rotemberg Versus
Calvo. The Economic Journal, 122(563):1115–1141.

Calvo, G. A. (1983). Staggered Prices in a Utility-Maximizing Framework. Journal of Monetary

Economics, 12(3):383–398.

Christiano, L. J., Eichenbaum, M., and Evans, C. L. (2005). Nominal Rigidities and the
Dynamic Effects of a Shock to Monetary Policy. Journal of Political Economy, 113(1):1–45.

Gali, J. (2015). Monetary Policy, Inflation, and the Business Cycle: an Introduction to the
New Keynesian Framework and its Applications.

Rotemberg, J. J. (1982). Monopolistic Price Adjustment and Aggregate Output. The Review
of Economic Studies, 49(4):517–531.

Smets, F. and Wouters, R. (2003). An Estimated Dynamic Stochastic General Equilibrium

Model of the Euro Area. Journal of the European Economic Association, 1(5):1123–1175.

Woodford, M. (2003). Interest and Prices: Foundations of a Theory of Monetary Policy.

Princeton University Press.

5
In general, the presence of monopolistic competition generates an inefficient low level of output in the
steady state of the New Keynesian model. In these lecture notes and in the lecture notes on the RBC model,
I normalize ex ante the steady-state level output to be equal to one. It turns out the parameter a in the NK
model of this lecture notes is lower than its counterpart in the lecture notes on the RBC model.

11