Prof. Dr.

Johannes Pfeifer Spring Term 2018

University of Cologne M.Sc. Program

ADVANCED MACROECONOMICS

-The Algebra of the RBC-Model-

1 Household problem
The household now maximizes the utility function
∞
!
X
t ct1−σ
max E0 β + ψ log(1 − lt ) + θgt (1)
{ct ,kt+1 ,lt }∞
t=0 t=0
1−σ

subject to the budget constraint

ct + it = wt lt + Rt kt − Tt ∀ t, (2)

the law of motion for capital with γx = 1 + n + g + ng

γx kt+1 = (1 − δ)kt + it ∀ t, (3)

an initial value for capital: k0 > 0, and the transversality/No-Ponzi constraint

lim E0 β t λt kt+1 = 0. (4)

t→∞

Additionally the household takes into account the stochastic laws of motion for zt

At = A∗ ezt (5)
 
zt = ρz zt−1 + εt , εt ∼ 0, σz2 (6)

and gt

gt = g ∗ eĝt (7)
 
ĝt = ρg ĝt−1 + εG G 2
t , εt ∼ 0, σG , (8)

and the government budget constraint

Tt = gt ∀ t (9)

Finally, we have non-negativity constraints

kt ≥ 0 (10)
ct ≥ 0 (11)
0 ≤ lt ≤ 1 . (12)

Due to our assumptions, we will not have corner solutions and can ignore the non-negativity
constraints.
This yields the Lagrangian
 1−σ 
∞
X  ct

+ ψ log(1 − lt ) + θgt


t
L = E0 β 1−σ
 
t=0 
−λt (ct + γx kt+1 − (1 − δ) kt + gt − wt lt − Rt kt )


1
with the First Order Conditions (∀t)
∂L  
= Et β t c−σ
t − λ t = 0 ⇒ ct
−σ
= λt (13)
∂ct
∂L 1 1
 
= Et β t −ψ + λ t wt = 0 ⇒ ψ = λ t wt (14)
∂lt 1 − lt 1 − lt
∂L β
 
= Et β t −λt + λt+1 ((1 − δ) + Rt+1 ) = 0
∂kt+1 γx
β
⇒ λt = Et λt+1 (Rt+1 + 1 − δ) . (15)
γx
and
∂L
= − (ct + γx kt+1 − (1 − δ) kt + gt − wt lt − Rt kt ) = 0, (16)
∂λt
that is the budget constraint.
The first three first order conditions can be combined to yield
β
c−σ
t = Et c−σ
t+1 (1 − δ + Rt+1 ) (17)
γx
1 wt
ψ = σ . (18)
1 − lt ct

2 Firm Problem
The firm problem is again static:
max A∗ ezt ktα lt1−α − wt lt − Rt kt
lt ,kt

The resulting first order conditions again state that factors are paid their marginal products
α
kt

wt = (1 − α)A∗ ezt (19)
lt
α−1
kt

∗ zt
Rt = αA e . (20)
lt

3 Market Clearing
Labor Market: labor supply (18) and demand (19) must be equal:
α
kt 1 σ

(1 − α)A∗ ezt =ψ c . (21)
lt 1 − lt t
Capital market: from (17) and (20) follows
( α−1 )
β kt+1

c−σ
t = Et c−σ ∗ zt+1
t+1 αA e + (1 − δ) . (22)
γx lt+1
Finally, the goods market equilibrium requires:
ct + γx kt+1 − (1 − δ)kt + gt = A∗ ezt ktα lt1−α . (23)

4 Approximating Technology
The Taylor approximation of technology At is given by

aex̂t ≈ a + a ex̂t  (x̂t − 0) = a + ax̂t ,

x̂t =0

where x̂t = zt (see Linearization Handout).

2
5 The Labor FOC
The labor FOC is given by
α
kt 1

(1 − α) A∗ ezt =ψ ct σ (24)
lt 1 − lt
The steady state is  α
k 1 σ
(1 − α) A∗ =ψ c . (25)
l 1−l
Loglinearizing yields
 α  2
k 1 1
 
∗
(1 − α) A zt + αk̂t − αˆlt = ψ σcσ ĉt − ψcσ (−1) lˆlt . (26)
l 1−l 1−l
Divide by the steady state
l ˆ
zt + αk̂t − αˆlt = σĉt + lt . (27)
1−l
Solving for l yields
−1 
l
 
ˆlt = +α zt + αk̂t − σĉt . (28)
1−l

6 The Euler Equation

The Euler Equation is given by
( α−1 !)
β kt+1

c−σ
t = Et c−σ
t+1 αAt+1 + (1 − δ) . (29)
γx lt+1

The steady state is ( )

 α−1
β k
1= αA∗ + (1 − δ) . (30)
γx l
Loglinearizing yields

  α−1 !
−σ −σ
 β −σ ∗ k
c − σc ĉt = Et 
γ c αA + (1 − δ)
 x l
| {z }
c−σ
( α−1 )
β k

∗
+ αA + (1 − δ) (−σ) c−σ ĉt+1
γx l
| {z }
1
(  α−1 )  #
β −σ k 
+ c αA∗ zt+1 + (α − 1) k̂t+1 − (α − 1) ˆlt+1 .
γx l

Subtracting the steady state and dividing by −σc−σ yields

"  α−1  #
β k 
ĉt = Et ĉt+1 − αA∗ zt+1 + (α − 1) k̂t+1 − (α − 1) ˆlt+1 . (31)
γx σ l

3
7 The Resource Constraint
The resource constraint is given by
ct + γx kt+1 − (1 − δ) kt + gt = At ktα lt1−α (32)
In steady state we have
c + γx k − (1 − δ) k + g = A∗ k α l1−α . (33)
Loglinearizing the single terms yields
ct ≈ c + cĉt (34)
γx kt+1 ≈ γx k + γx k k̂t+1 (35)
(1 − δ) kt ≈ (1 − δ) k + (1 − δ) k k̂t (36)
∗ α 1−α ∗ α 1−α ∗ α 1−α
At ktα lt1−α ≈A k l +A k l zt + αA k l k̂t + (1 − α) A∗ k α l1−α ˆlt (37)
ĝt
ge ≈ g + gĝt . (38)
Together
c + cĉt + γx k + γx k k̂t+1 − (1 − δ) k − (1 − δ) k k̂t + g + gĝt = (39)
A∗ k α l1−α + A∗ k α l1−α zt + αA∗ k α l1−α k̂t + (1 − α) A∗ k α l1−α ˆlt .
Rewrite as
c + γx k − (1 − δ) k + g +cĉt + γx k k̂t+1 − (1 − δ) k k̂t + gĝt =
| {z }
A∗ kα l1−α
 
A∗ k α l1−α + A∗ k α l1−α zt + αk̂t + (1 − α) ˆlt . (40)

Solving for k̂t yields

cĉt = A∗ k α l1−α zt + αA∗ k α l1−α k̂t + (1 − α) A∗ k α l1−α ˆlt − γx k k̂t+1 + (1 − δ) k k̂t − gĝt
 
cĉt = [αy + (1 − δ) k] k̂t + y zt + (1 − α) ˆlt − γx k k̂t+1 − gĝt
1 y y  c g
  
k̂t+1 = α + (1 − δ) k̂t + zt + (1 − α) ˆlt − ĉt − ĝt . (41)
γx k γx k γx k γx k

8 Eliminating l
Write (28) as
−1 
l
   
ˆlt = +α zt + αk̂t − σĉt ≡ γl zt + αk̂t − σĉt (42)
1−l
and plug in into (41)
1 y y y c g
   
k̂t+1 = α + (1 − δ) k̂t + zt + (1 − α) γl zt + αk̂t − σĉt − ĉt − ĝt
γx k γx k γx k γx k γx k
1 y y y y
   
k̂t+1 = α + (1 − δ) + (1 − α) γl α k̂t + + (1 − α) γl zt
γx k k γx k γx k
c y g
 
− + (1 − α) γl σ ĉt − ĝt
γx k γx k γx k
1 y c y
   
k̂t+1 = α (1 + (1 − α) γl ) + (1 − δ) k̂t + − + (1 − α) γl σ ĉt
γx k γx k γx k
| {z } | {z }
α1 α2
y g
 
+ (1 + (1 − α) γl ) zt + − ĝt (43)
γx k γx k
| {z } | {z }
α3 α4

4
and into (31)
"  α−1  #
β k  
ĉt = Et ĉt+1 − αA∗ zt+1 + (α − 1) k̂t+1 − (α − 1) γl zt+1 + αk̂t+1 − σĉt+1
γx σ l
  α−1 !  α−1 
β k β k
 1− αA∗ (α − 1) γl σ ct+1 − αA∗ (1 − (α − 1) γl ) zt+1 
 γ x σ l γx σ l 
ĉt = Et  
 α−1


β k
αA∗
 
− ((α − 1) − γl (α − 1) α) k̂t+1
γx σ l
  α−1  α−1 ! 
β k β k
− αA∗ (α − 1) (1 − γl α) k̂t+1 + 1 − αA∗ (α − 1) γl ct+1 +
 γx σ l γx l 
| {z } | {z } 
 α5 α6

ĉt = Et   .
 
 α−1
 β ∗ k

− αA (1 − (α − 1) γl ) zt+1
 
 γx σ l


| {z }
α7
(44)

9 Method of Undetermined Coefficients

Guess a solution of the form

k̂t+1 = φkk k̂t + φkz zt + φkg ĝt (45)

ĉt = φck k̂t + φcz zt + φcg ĝt . (46)

Plugging in Equation (43) yields

k̂t+1 = α1 k̂t + α2 ĉt + α3 zt + α4 ĝt

 
φkk k̂t + φkz zt + φkg ĝt = α1 k̂t + α2 φck k̂t + φcz zt + φcg ĝt + α3 zt + α4 ĝt .

This equation has to hold for all values of the state variables ĝt , k̂t , zt . In turn setting all states
to 0 except for one, which is set to 1, yields the following equations (this is equivalent to simply
comparing the sides):

φkk = α1 + α2 φck (47)

φkz = α3 + α2 φcz (48)
φkg = α4 + α2 φcg (49)

Equation (44):
h i
ĉt = Et α5 k̂t+1 + α6 ĉt+1 + α7 zt+1
h   i
φck k̂t + φcz zt + φcg ĝt = Et α5 k̂t+1 + α6 φck k̂t+1 + φcz zt+1 + φcg ĝt+1 + α7 zt+1
h i
φck k̂t + φcz zt + φcg ĝt = Et (α6 φcz + α7 ) zt+1 + (α6 φck + α5 ) k̂t+1 + α6 φcg ĝt+1

Plugging in the laws of motion for zt+1 (6) and gt+1 (8) and using the expectations operator
yields:

5
  
(α6 φcz + α7 ) (ρzt + εt+1 )
  
φck k̂t + φcz zt + φcg ĝt =Et + (α6 φck + α5 ) φkk k̂t + φkz zt + φkg ĝt 
 
   
+α6 φcg ρg gt + εG
t+1

φck k̂t + φcz zt + φcg ĝt = (ρ (α6 φcz + α7 ) + (α6 φck + α5 ) φkz ) zt
+ (α6 φck + α5 ) φkk k̂t
+ ((α6 φck + α5 ) φkg + α6 φcg ρg ) ĝt .
Comparing sides or in turn setting all states to 0 except for one, which is set to 1, yields:

φck = (α6 φck + α5 ) φkk (50)

φcz = ρ (α6 φcz + α7 ) + (α6 φck + α5 ) φkz (51)
φcg = (α6 φck + α5 ) φkg + α6 φcg ρg . (52)
We now have 6 equations in six unknown coefficients φ. Hence, we need to solve this system.
Use (47) and (50)

φkk α1
φkk = α1 + α2 φck ⇒ φck = − (53)
α2 α2
φck = (α6 φck + α5 ) φkk . (50)
Set (53) and (50) equal and transform to get quadratic equation
φkk α1 φkk α1 α6 2 α6 α1
     
− = α6 − + α5 φkk = φkk + α5 − φkk
α2 α2 α2 α2 α2 α2
α6 2 α6 α1 1 α1
 
φkk + α5 − − φkk + =0
α2 α2 α2 α2
α5 α2 α6 α1 1 α1
 
φ2kk + − − φkk + =0 .
α6 α6 α6 α6
The solution for φkk is given as
s 2
1 − α5 α2 + α6 α1 1 − α5 α2 + α6 α1 α1
 
φkk = ± − . (54)
2α6 2α6 α6
φck then follows from:
φkk α1
φck = − . (53)
α2 α2
Second, use
φkz = α3 + α2 φcz (48)
φcz = ρ (α6 φcz + α7 ) + (α6 φck + α5 ) φkz (51)
to get

φkz = α3 + α2 φcz (55)

φcz = ρ (α6 φcz + α7 ) + (α6 φck + α5 ) φkz (56)
φcz = α6 ρφcz + ρα7 + α6 φck α3 + α6 φck α2 φcz + α3 α5 + α5 α2 φcz
(57)
φcz (1 − α6 ρ − α6 φck α2 − α5 α2 ) = ρα7 + α6 φck α3 + α3 α5 (58)
ρα7 + α6 φck α3 + α3 α5
φcz = . (59)
(1 − α6 ρ − α6 φck α2 − α5 α2 )

6
Given φcz , φkz can be computed from

φkz = α3 + α2 φcz . (48)

Finally, plug

φkg = α4 + α2 φcg (49)

into
φcg = (α6 φck + α5 ) φkg + α6 φcg ρg (52)
to get

φcg = (α6 φck + α5 ) (α4 + α2 φcg ) + α6 φcg ρg

(α6 φck + α5 ) α4
φcg = . (60)
1 − (α6 φck + α5 ) α2 − α6 ρg