Derivation and Log-linearization of Otsu

(2007)s Small Open Economy Model

Otsu, Keisuke (2007), "A Neoclassical Analysis of the Asian Crisis: Business Cycle Accounting of
a Small Open Economy," IMES Discussion Paper Series No. 07-E-16, Bank of Japan.

The benchmark prototype economy

max U = E0

1
X

t u( c ; l )
t t

t=0

The utility is:

u(ct; lt) = log(ct

lt )

where parameters (> 0) and (> 1) represent the level and curvature of the
utility cost of labor respectively.
subject to the budget constraint:

wt
lt

lt+r tkt+tt+

dt+1
= ct+ xtxt+dt+ (dt+1)
R dt

The law of motion for capital:

kt+1 = xt + (1

)kt

where dene = (1 + )(1 + n), is the growth rate of labor augmenting

The debt adjustment cost function,

(dt+1 d)2
,
2

(dt+1), as

Solution of consumersmaximization problem:

max

fct;lt;kt+1 ;dt+1 g

U = E0

1
X

t=0

t log (c

lt )

where d is the

wt

dt+1
lt + rtkt + tt +
= ct + xt [ kt+1
R
lt
dt

(1

)kt] + dt +

(dt+1)

The Lagrangian

L =

1
X

t=0
1
X

t=0

t log(c

8
<
:

lt ) +
d
wt
lt + rtkt + tt + R t+1
lt
dt
(1
)kt] dt
xt [ kt+1

ct

9
=

(dt+1) ;

First-order-condition of the model show in Equation (A1) to (A4) can be simplied to the below four equations.

@L
= tuct
@ct

=0

@L
wt
= tult + t t
=0
@lt
lt
@L
=
@kt+1

xt

+ t+1 t+1[ t+1 + xt+1(1

"

@L
= t t
@dt+1
R dt

(dt+1)

1
ct

lt

= t

t+1

t+1

)] = 0

=0

(A1)

lt 1
=
ct
lt

wt
lt

Use Equation (A1) substitute t, and we obtain:

lt
t xt

1
ct

lt

xt

wt

(A2)

lt

= t+1 [ t+1 + xt+1(1

)]

= Etuct+1[ t+1 + xt+1(1

)]

(A3)

"

1
ct

lt

R dt
"

(dt+1) = t+1
0

R dt

(dt+1) = Etuct+1

(A4)

The rm produces a nal good (output y with a constant returns to scale) from
capital and labor using a Cobb-Douglas production function.

yt = ztkt lt1

The rms prot maximization problem is:

= yt

w t lt

r t kt

Solution of rmsmaximization problem:

max

fkt;ltg

= ztkt lt1

@
= ztkt 1lt1
@kt
@
= (1
@lt

)z t k t lt

w t lt

r t kt

rt = 0
wt = 0

ztkt 1lt1

= rt

yt
= rt
kt

(1

)z t k t lt
(1

(A5)
= wt

yt
) = wt
lt

(A6)

Substitute (A6) into (A2), and we get the labor wedge.

wt
lt 1 =
= (1
lt

yt 1
lt lt

lt

= (1

yt 1
)
lt lt

Substitute (A5) into (A3), and we get the investment wedge.

1
xt

Substitute
wedge.

ct
0 (d
t+1 )

lt

"

= Et uct+1

(dt+1

yt+1
+ xt+1(1
kt+1

!#

d) into (A4), and we get the foreign debt

1
ct

lt

"

(dt+1

R dt

d) = Et[uct+1]

The government collects tax revenues at date t, and the government budget
constraint:

tt = (1

)wtlt + ( xt

1)xt

lt

Substitute the government budget constraint into the consumer budget constraint and we can get the economys resource constraint at last:

wt
lt

lt + rtkt + (1

)wtlt + ( xt

1)xt +

lt

r t k t + w t lt

dt+1
xt +
= ct + dt +
R dt

(dt+1)

dt+1
+
R dt

(dt+1)

rtkt + wtlt = xt + ct + dt

yt = ct + xt + dt

where the trade balance is:

dt+1
= ct + xtxt + dt + (dt+1)
R dt

(dt+1
dt+1
+
R dt
2

d)2

tbt = dt

dt+1
(dt+1
+
R dt
2

d)2

The equilibrium conditions in the detrended per-capita form:

yt = ct + xt + dt

(dt+1
dt+1
+
R dt
2

yt = ztkt lt1
lt 1 = (1

d)2

(1)
(2)

yt 1
lt lt

(3)

1
xt

ct

lt

1
ct

lt

"

= Et uct+1
"

1
R dt

(dt+1

yt+1
+ xt+1(1
kt+1
#

d) = Et[uct+1]

Endogenous variables are : ct ,xt ,yt ,kt ,dt ,lt.

Exogenous variables are: zt, dt, xt, and lt.

We need the capital accumulation law to close the model.

!#

(4)

(5)

kt+1 = xt + (1

)kt

Steady state of the prototype model

1
ct

lt

"

1
R dt

(dt+1

d) = Et[uct+1]

(6)

1
c

[
l R d

(d
1

d)] = uc

(ss1)

=R

From equation(4), the calculation process is below:

1
xt

ct

lt
x

"

yt+1
+ xt+1(1
kt+1

= Et uct+1

"

k
z
l

+ x(1

!#

=z
1

k
l
k
=
l

1
z

1
=
z
"

k
l

+ x(1
"

x (1

x (1

)
#)

From equation(6), the calculation process is below:

kt+1 = xt + (1

)kt

1
1

(ss2)

k = x + (1
x=[

)k

(1

(ss3)

)]k

From equation(2), the calculation process is below:

yt = ztkt lt1
y=

zk l1

k
=z
l

k=z

From equation(1), the calculation process is below:

k
l

yt = ct + xt + dt

dt+1
(dt+1
+
R dt
2
d

y =c+x+d

c=y

d+

R d

k
c=z
l
c=z

d)2

(d

k
l

R d

(1

2
1

k
=z
l

(1

d)2

(d

d)2

k
k
)]
l
l

)]

k
l
l

(1

d+

)]k
d

R d

d+ d

d+

d
R d

c=z

k
l

(1

k
l+(
l

)]

From equation(3), the calculation process is below:

lt 1 = (1
l

= (1

l = (1

yt 1
lt lt

y1
l l

k
)z
l

1
l

1)d

(ss4)

l = (1
"

l = (1

k
l
k
l

1
l

#1

(ss5)

We obtain kl rst, and in order obtain l, then k, then c, then y , then x.

k
k= l
l
k
c=z
l

(1

k
)]
l + d(
l

1)

y=z
x=[

k
l

(1
d=d

Log-linearization
c^t = log ct

log c

x
^t = log xt

log x

)]k

y^t = log yt

log y

^t = log kt
k

log k

^
lt = log lt

log l

z^t = log zt
d^t = ddt

log z

^dt = log dt

log d

^xt = log xt

^lt = log lt

log x

log l

Equilibrium conditions in the detrended per-capita form:

yt = ct + xt + dt

(dt+1
dt+1
+
R dt
2

yt = ztkt lt1

d)2

(1)
(2)

lt

1
xt

ct

lt

1
ct

lt

= (1

"

= Et uct+1
"

1
R dt

yt 1
)
lt lt
yt+1
+ xt+1(1
kt+1

(dt+1

kt+1 = xt + (1

From equation (1)

d) = Et[uct+1]

)kt

(3)

!#

(4)

(5)
(6)

dt+1
(dt+1
+
R dt
2

yt = ct + xt + dt

yey^t

cec^t

+ xex^t

+ d^t + 1 d

d^t+1 + 1 d
R de^dt

y (1+^
yt) = c (1 + c^t)+x(1+^
xt)+ d^t + 1 d

d)2
h

d^t+1d + d
R d(1 + ^dt)

i2

d^t+1 + 1 d

d^t+1d + d
^
y + y y^t = c + cc^t + x + xx
^t + ddt + d
+
R d(1 + ^dt)

d^t+1d + d

2
2
d^t+1d

where y = c + x + d

d
R d

and

d^t+1 d

^t+1d + d
d
y + y y^t = c + cc^t + x + xx
^t + dd^t + d
R d(1 + ^dt)

d
R d

d
R d

R d(1 + ^dt)y y^t = R d(1 + ^dt)cc^t + R d(1 + ^dt)xx

dd^t+1

d + d + d^dt

c
x
d
d ^
d
y^t = c^t + x
^t + d^t
dt+1 +
^dt
y
y
y
R dy
R dy

(1.c)

d^t+1d

equation (2)

yt = ztkt lt1
^

yey^t = zez^t (kekt ) (lelt )1

^
)ez^t+ kt+(1

= (zk l1

^
ey^t = ez^t+ kt+(1

)^
lt

^t + (1
(1 + y^t) = 1 + z^t + k
^t + (1
y^t = z^t + k
Equation (3)

)^
lt

)^
lt

)^
lt
(2.c)

lt

l
1

l
l

1+(

1+(

)zez^t

= (1
1)^
lt

1 e(
1 e(

kt
)zt
lt

= (1

^
lelt

= (1

1)^
lt+^lt

^
kekt

^
)zez^t k e kt l

= (1

1) ^
lt + ^lt = (1
i

1) ^
lt + ^lt = (1

1
lt

^
lt

^
)zez^t k e kt l

)zk l
)zk l1

1
^
l e lt

^
lelt

1
^
l e lt
e

^
lt

^t
1 + z^t + k
^t
1 + z^t + k

^
lt
^
lt

where

(1

= (1

)y 1 + (
1+(

)y and y = zk l1

1) ^
lt + ^lt = (1

^t
1) ^
lt + ^lt = 1 + z^t + k
^t
^lt = z^t + k

^
lt

^t + (1
^lt = z^t + k

equation (4)

^t
)y 1 + z^t + k

^
lt

^
lt

1) ^
lt
)^
lt

(3.c)

1
xt

e^xt

ct

lt

1
cec^t

l e ^lt

"

yt+1
+ xt+1(1
kt+1

= Et uct+1
2

= Et 4

1
^
l e lt+1

cec^t+1

!#

^
yey^t+1 k 1e kt+1
)
+ xe^xt+1 (1

!3
5

Ignore Et for a moment

^
x e xt

cec^t+1

^
l e lt+1

^
yey^t+1 k 1e kt+1 + xe^xt+1 (1

cec^t

^
l e lt

LHS
^
c^
x e xt ce t+1

= x ce^xt+^ct+1

^
x e xt
x

= x c (1 + ^xt + c^t+1)

^
l e lt+1
^
l e^xt+ lt+1

1 + ^xt + ^
lt+1

= x c + x c^xt + x cc^t+1

= (c

RHS

l ) x ^xt + (c

l )

+ x cc^t+1

l ^xt
x

^
lt+1

^
lt+1

^
yey^t+1 k 1e kt+1 + xe^xt+1 (1

2
6
6
4

^
yey^t+1 k 1e kt+1 cec^t

2
6
6
4
2
4

^
x e xt+1 (1

+ (1
c ky

+(1

)c xe^xt+1+^ct
1 + y^t+1

^
l e lt

^
^
yey^t+1 k 1e kt+1 l e lt
^
) l e lt

^
x e xt+1 (1

)cec^t

^
c ky ey^t+1 kt+1+^ct

cec^t

(1

^t+1 + c^t
k

)c x (1 + ^xt+1 + c^t)

^t+1 + ^
y y^t+1 k
lt
e
k
^xt+1 + ^
lt
e
x

) l

(1

y
k

1 + y^t+1
) l

7
7
5

7
7
5

^t+1 + ^
k
lt

lt
x 1 + ^ xt+1 + ^

3
5

=
2

(c

=4

(1

(c

(c

l )

y
k

) (c

(c
(1

=4
=

y
^t+1 + c y c^t
^t+1
1 + y^t+1 k
l
^t+1 k
k
k 1+y
l ky ^
lt + (1
)c x (1 + ^xt+1) + (1
)c xc^t
(1
) l x (1 + ^xt+1) (1
lt
) x l ^

c ky
6
6
4

l )

l )

1 + y^t+1

^t+1 +
k

l ) x (1 + ^xt+1)

y
+ (1
h k
y
k + (1

"

y
k

1 + y^t+1

+(1

y
k

+
)

^t+1 + (1
k

y
^t+1
ky
x + (1

7
7
5

) x cc^t
5
i
^
) x l lt

^t+1 + x cc^t
l ) ky 1 + y^t+1 k
5
) (c
l ^
lt
l ) x (1 + ^xt+1)
x

) x (1 + ^xt+1) + x cc^t

y^
k kt+1
) x ^xt+1

+ x cc^t

^
lt

^
lt

= (c

= (c

l ) x +

l )

where

(c
0
@
h

l )

"

(c
+ (c

y
k

y
^t+1
ky

+(1

l )
l ) (1

+ x(1

y^
k kt+1
x ^ xt+1

^t
x cc

y
y^
y
^
(
c
l
)
t+1
k
k kt+1
x
) x ^xt+1 + x cc^t
l

) and

= ky + x(1

^
lt

^
lt
1
A

LHS=RHS
(c
= (c

l ) x ^xt + (c
l )

x+

0
@

l )

(c
+ (c

+ x cc^t+1

l )
l ) (1

^
lt+1

y
y^
y
^
(
c
l
)
k t+1
k kt+1
x
) x ^xt+1 + x cc^t
l

^
lt

1
A

(c

l )

x ^ xt

0
@

(c
+ (c

^t+1
x cc

x ^ xt

0
@

(c

y^
y
y
^
(
c
l
)
t+1
k
k kt+1
x
l
) x ^xt+1 + x cc^t

l )
l ) (1

+ x

^
lt+1

y
^t+1
ky

y^
k kt+1
x
cc^
(c l ) t

x c

l )

c^t+1 +

(c

+ (1
(c

l
l )

) x ^xt+1
A
^
l lt
l )

^
lt+1

^
lt

1
A

^xt =

y
y^
(1
)
c
c^t
y^t+1
kt+1 +
^xt+1 +
(c
l )
xk
xk
c
l
^
c^t+1 +
lt+1
(c
l )
(c
l )

(c

l )

Add back expectation

^xt =

y
Et y^t+1
xk
l
^
lt
l )
(c

^t+1 +
k

(c

lt+1 +
Et^

l )

(1

ct
(^

Etc^t+1)

)Et ^xt+1

(4.c)

^
lt

equation (5)

1
ct

1
cec^t

lt

"

1
R dt

"

1
l e ^lt R de^dt

Ignore Et for the moment.

(dt+1

d) = Et[uct+1]

d^t+1 + 1 d

= Et

1
cec^t+1

^
l e lt+1

^
l e lt+1

cec^t+1

1
^
d e dt

cec^t+1 R

1
R e^dt
d

d^t+1d + d

^
cec^t+1 dd^t+1+ l e lt+1 R

cec^t

1 + l e ^
lt+1 dd^
t+1
^
d e dt

^
l e lt

LHS
cec^t+1 R

1
^
d e dt

= c ec^t+1 ^dt R
d
2

=4

^
cec^t+1 dd^t+1 + l e lt+1 R

^
+ l e lt+1 dd^t+1

^
^
cec^t+1 dd^t+1 + l e lt+1 ^dt R + l e lt+1 dd^t+1
d

(1 + c^t+1
d
l
1+ ^
lt+1
d

1
^
d e dt

^dt) R
^dt R

c (1 + c^t+1) dd^t+1
5
+ l 1+ ^
lt+1 dd^t+1

cec^t

^
l e lt

=4

l
d

^dt R c (1 + c^t+1)
(1 + c^t+1) R
d
1+ ^
lt+1 R + l R ^dt + l 1 + ^
lt+1
d

where R 1 =
d
= c (1 + c^t+1)
=
=

c + cc^t+1

(c

RHS

d^t+1
dd^t+1

3
5

or R d = 1
c^dt
c^dt

l ) + cc^t+1

c dd^t+1
1 c dd^
t+1

(c

l ) ^dt

l
l

1+ ^
lt+1 +
l
1 (c

^
lt+1 + l ^dt +
l ) dd^t+1

dd^t+1

l ^dt + l
1 l

dd^t+1

^
lt+1

^
l e lt

cec^t

c (1 + c^t)

c + cc^t

l
l

1+ ^
lt
l

^
lt

LHS=RHS
"

(c

cc^t+1

l ) + cc^t+1 (c
1 (c
l ) dd^t+1

(c

l ) ^dt

1 (c

l ) ^dt
l ^
lt+1

l ) dd^t+1

c + cc^t

^
lt+1 = cc^t

^
lt

^
lt

(c

l ) ^dt = cc^t+1

^dt = (c c l ) (^
ct+1

cc^t
c^t)

l
l
(c

^
lt+1 + l
^
l
l ) t+1

^
lt

^
lt

1 (c

l ) dd^t+1

1 dd^
t+1

Add back expectation.

^dt =

(c

l )

equation (6)

(Etc^t+1

c^t)

(c

l )

Et^
lt+1

^
lt

dEtd^t+1

(5.c)

kt+1 = xt + (1
^

)kt
^

kekt+1 = xex^t + (1
)kekt
^t+1) = x(1 + x
^t)
k(1 + k
^t) + (1
)k(1 + k
^t+1 = xx
^t
kk
^t + (1
)k k
x
^
^t + (1
kt+1 = x
k

^t
)k

(6.c)

We will solve the model by the linearizing the equations that characterize the
competitive equilibrium around the steady state.

Here again the log-linearized model :

c
x
d
d ^
d
y^t = c^t + x
^t + d^t
dt+1 +
^dt
y
y
y
R dy
R dy

^t + (1
y^t = z^t + k

)^
lt

^t + (1
^lt = z^t + k
^xt =

^dt =

(c

l )

y
Et y^t+1
xk
l
^
lt
(c
l )

(Etc^t+1

c^t)

^t+1 +
k

(c

l )

(2.c)

)^
lt
c

(c

Et^
lt+1 +
l

(1.c)

l )

(1

Et^
lt+1

(3.c)
(^
ct

Etc^t+1)

(4.c)

)Et ^xt+1
^
lt

dEtd^t+1

(5.c)

x
^
^t + (1
kt+1 = x
k

^t
)k

(6.c)