Beruflich Dokumente
Kultur Dokumente
August 2011
Contents
A Appendix: Model Solution 2
A.1 Equilibrium Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
A.1.1 Country-Specific Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
A.1.2 Bilateral Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
A.1.3 Important Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
A.1.4 Relationships Between Model Variables and Observed Data . . . . . . . . . . . . . 15
A.1.5 Decomposition of Marginal Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
A.2 Balanced Growth Path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
A.2.1 Calibrated Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
A.2.2 Composite Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
B Appendix: Data 24
1
A Appendix: Model Solution
A.1 Equilibrium Conditions
This appendix summarizes the equilibrium conditions of the model both in nonlinear and linearized form.
The stochastic processes for the exogenous variables are presented in Table 1, in the main body of the
paper.
Xi,t
∗
refers to the value of variable Xi at time t along the balanced growth path. Xi,0
∗
is the initial
value of the variable at time 0. x̂i,t denotes the log-deviation of variable Xi at time t from its value along
the balanced growth path. xi,t denotes the absolute deviation of variable Xi at time t from its value along
the balanced growth path.
−1 C
P1,t
c
Z1,t C1,t − κ1 C1,t−1 = λq1,t d
, (1)
P1,t
L∗1,0 q
f
ŵ1,t = χ1 ˆl1,t − λ̂1,t . (4)
1 − L1,0
∗
f d
Note, that w1,t is the desired real wage expressed in terms of P1,t .
2
(a) If wages are flexible:
f θw∗
1,0 w
w1,t = w1,t + w∗ θ̂ 1,t (5)
1 + θ1,0
r W1,t (h)
The relative wage is defined as W1,t = W1,t
. The linearized equation is given by:
1 β1
(ω 1,t − ιw w
1 ω 1,t−1 ) − ∗ (ω 1,t+1 − ι1 ω 1,t )
π1
∗
π1
w
(1 − ξ w1 β 1 ) (1 − ξ 1 ) f θw
1 w
= ŵ1,t+j − ŵ1,t+j + θ̂1,t (7)
ξw1 1 + θw
1
1
As customary, in the numerical implementation we using the approximation π ∗1
(ω 1,t − ιw
1 ω 1,t−1 ) =
ω 1,t − ιw
1 ω 1,t−1 . See for example Smets and Wouters (2007).
i
q̂1,t + ẑ1,t = ψ i1 µ2z (ı̂1,t − ı̂1,t−1 ) − β 1 ψ i1 µ2z (ı̂1,t+1 − ı̂1,t ) . (11)
3
6. First order condition for K1,t :
d
P1,t λq1,t+1 R1,t+1
k i
P1,t+1 d
P1,t λq1,t+1
q1,t = β 1 + (1 − δ 1 )β q
1 1,t+1 d . (12)
i
P1,t λq1,t P1,t+1
d i
P1,t+1 P1,t λq1,t
7. Capital accumulation:
2 !
ψi I1,t
i
K1,t = (1 − δ 1 ) K1,t−1 + Z1,t I1,t 1− 1 − µz . (14)
2 I1,t−1
8. Consumption basket:
ρo ρo
1+ρo1
1 1+ρ
1 1 1+ρ
1
(ω cc 1+ρo ne
(ω oc µtzo Z1,t
o c
o 1+ρo o
C1,t = 1 )
1 C1,t 1 + 1 )
1 O1,t 1 . (16)
ĉ1,t = ω cc ĉ
1 1,t
ne
+ ω oc
1 o c
1,t + ẑ o
1,t . (17)
c
ĉne c d mc
1,t = ω 1 ĉ1,t + ω 1
m
m̂1,t + ẑ1,t . (19)
4
d
10. First order condition for C1,t in consumption basket:
c 1
1+ρ
ρo
ne 1+ρ
1 ρc
P1,t ω cc
1 C1,t
o
1 ω c1 C1,t c
1
d ne d
= 1. (20)
P1,t C1,t C1,t
c
11. First order condition for M1,t in consumption basket:
m c 1+ρ
1
ρo 1+ρ
ρc
1
P1,t P1,t C1,t o
1 ω mc ne
1 C1,t
c
1
d
= d
ω cc
1 ne m c
m
Z1,t . (22)
P1,t P1,t C1,t Z1,t M1,t
c
12. First order condition for O1,t in consumption basket:
o c 1+ρ
ρo
1
P1,t P1,t ω oc
1 C1,t
o
1
d
= d c c
µtzo Z1,t
c
. (24)
P1,t P1,t µtzo Z1,t O1,t
i
ı̂1,t = ω i1 ı̂d1,t + ω mi
1
m
m1,t + ẑ1,t . (27)
5
14. First order condition for Itd in investment basket:
ρc
1+ρ
1
i
P1,t ω i1 I1,t c
1
d d
= 1. (28)
P1,t I1,t
i
15. First order condition for M12,t in investment basket:
ρc
1+ρ
1
m i
P1,t P1,t ω mi
1 I1,t
c
1
m
d
= d m i
Z1,t . (30)
P1,t P1,t Z1,t M1,t
with
1+ρ
1
K1,0
∗ v
1
φk1 = ω k1 , (34)
ω k1 µz V1,0
∗
1+ρ
1
L∗1,0 v
1
φl1 = ω l1 , (35)
ω l1 V1,0
∗
6
17. Output aggregator:
ρo ρo
1+ρo1
1
vy 1+ρo
1
o
1
oy 1+ρo t o y
1+ρ
1
o
Y1,t = (ω 1 ) 1 (V1,t ) 1 + (ω 1 ) 1 µzo Z1,t O1,t
1+ρ 1 . (36)
1+ρ ρv
1
ρo 1+ρ
1
k
R1,t M C1,t Y1,t 1
o
ω k1 V1,t v
1
d
= d
ω vy
1 . (38)
P1,t P1,t V1,t K1,t−1
k ρo1 ρv1
r̂1,t c 1,t +
= mc (ŷ 1,t − v̂ 1,t ) + v̂ 1,t − k̂1,t−1 . (39)
1 + ρo1 1 + ρv1
1+ρ ρv
1
ρo 1+ρ
1
1 W1,t M C1,t ω vy
1 Y1,t
o
1 ω l1 V1,t v
1
d
= d
. (40)
µtz Z1,t P1,t P1,t V1,t µtz Z1,t L1,t
ρo1 ρv1 ˆ
ŵ1,t c 1,t +
= mc (ŷ1,t − v̂1,t ) + v̂1,t − ẑ1,t − l1,t + ẑ1,t . (41)
1 + ρo1 1 + ρv1
y
20. First order condition for O1,t in output aggregator:
1+ρ
ρo
1
o
P1,t M C1,t ω oy
1 Y1,t
o
1
d
= d t o y µtzo Z1,t
o
. (42)
P1,t P1,t µzo Z1,t O1,t
7
(a) If prices are flexible:
M C1,t 1 + τ p1
= . (44)
d
P1,t 1 + θp1,t
The linearized equation is given by:
1
c 1,t = −
mc θˆp 1,t . (45)
1 + θp∗
1,0
(b) If prices are sticky and export prices are set in the currency of the producer:
X ∞
r p j 1 + τ p1 l d
P1,t (i) (ξ 1 ) ψ 1,t,t+j p π 1,t,j Y1,t+j (i)
j=0
θ 1,t+j
"∞ p
#
X p P d
M C1,t+j 1,t+j d 1 + θ 1,t+j
= (ξ 1 ) ψ 1,t,t+j d d
Y1,t+j (i) p , (46)
j=0
P1,t+j P1,t θ1,t+j
where
d
r
P1,t (i)
P1,t = d
(47)
P1,t
d
P1,t λc1,t+j
ψ 1,t,t+j = β j1 (48)
d
P1,t+j λc1,t
j
Y ι
π l1,t,j = π d1,t−1+i (π ∗1 )1−ι (49)
i=1
p
!− 1+θp 1,t+j
d d l
d d
P1,t (i) P1,t π 1,t,j θ
1,t+j
Y1,t+j (i) = Y1,t+j d d
(50)
P1,t P1,t+j
or after simplifying
!−1
X
∞
1+ τ p1 M C1,t+j d l
P1,t π 1,t,j 1+ θp1,t+j
Ξ1,t+j p
r
P1,t − =0
j=0
θ1,t+j d
P1,t+1 d
P1,t+j θp1,t+j
!− p
1
1+θ
p
c d l
p j λ1,t+j P1,t π 1,t,j θ
1,t+j
r
− θ
p
1,t+j
d
Ξ1,t+j = (ξ 1 β 1 ) P1,t 1,t+j Y1,t+j . (51)
λc1,t d
P1,t+j
The linearized equation is given by
1 d p d
β1 d p d
(1 − ξ p1 β 1 ) (1 − ξ p1 ) θp∗
1,0 p
π − ι1 π t−1 = ∗ π t+1 − ι1 π t + c 1,t +
mc p∗ θ̂ 1,t (52)
π ∗1 1,t π1 ξ p1 1 + θ1,0
In the numerical implementation we use the approximation typically used in the literature
p d
π
1
∗ π d
1,t − ι1 π t−1 = π d1,t − ιp1 π dt−1 . See, for example, Smets and Wouters (2007).
1
8
22. Government spending Gd1,t :
g
Gd1,t = g1 Z1,t d
Y1,t . (53)
d d g
ĝ1,t = ŷ1,t + ẑ1,t . (54)
d
23. Market clearing condition for Y1,t :
d d d
Y1,t = I1,t + C1,t + Gd1,t + X1,t . (55)
c y
O1,t = O1,t + O1,t . (57)
s rs
25. Nominal interest rate R1,t and real interest rate R1,t :
1 λq1,t+1 P1,t
d
= β . (59)
s
1 + R1,t λq1,t P1,t+1
d
s rs
assuming that β 1 /µz is close to 1. r1,t and r1,t are measured in absolute deviation from their values
along the balanced growth path.
9
s
26. Monetary policy reaction function i1,t = R1,t − 1:
y gap
i1,t = ī1 + γ i1 (i1,t−1 − ī1 ) + (1 − γ i1 )(π core π core core
1,t + γ 1 (π 1,t − π̄ 1,t ) + γ 1 y1,t ). (61)
s y gap
r1,t = γ i1 r1,t−1
s
+ (1 − γ i1 )(π core π core core
1,t + γ 1 (π 1,t − π̄ 1,t ) + γ 1 y1,t ). (62)
ne
27. Core price level P1,t :
1+ρ
ρo
1
ne c ω cc
1 C1,t
o
1
P1,t = P1,t ne
. (63)
C1,t
In constructing the core price index the shock to oil efficiency enters, as this shock changes the
c
share of oil in the headline price index P1,t .
10
30. Inflation of headline prices:
c
head
P1,t
π 1,t = log c
. (68)
P1,t−1
1 c i
X1,t = M2,t + M2,t , (72)
ζ1
c i
as country 1’s real per capita exports X1,t and country 2’s real per capita imports M2,t + M2,t are
related through ζ 1 . The linearized equation is given by:
c∗ i∗
M2,0 c
M2,0
x̂1,t = c∗ i∗
m̂2,t + c∗ i∗
m̂i2,t . (73)
M2,0 + M2,0 M2,0 + M2,0
11
34. Nonoil trade balance to gross output ratio:
λq2,t+1 P2,t
d c
P2,t+1 λq Pd Pc
b rer1,t+1 1,t+1 1,t 1,t+1
= φ1,t . (81)
λq2,t P2,t
c d
P2,t+1 rer1,t λq1,t P1,t
c d
P1,t+1
12
37. Net foreign asset condition:
b
e1,t P2,t B1,t bal
= e1,t B1,t−1 + N T1,t . (84)
φb1,t
e1,t B1,t−1
where b1,t−1 is the absolute deviation of d Yd
P1,t
from 0. tbal
1,t is the deviation of the trade balance
1,t
to gross output ratio from its value along the balanced growth path.
o o
ζ 1 Y1,t + Y2,t = ζ 1 O1,t + O2,t . (86)
o∗
ŷ
o∗ 1,t
+ o∗
ŷ
o∗ 2,t
= ô 1,t + ô2,t . (87)
ζ 1 Y1,0 + Y2,0 ζ 1 Y1,0 + Y2,0 ζ 1 O1,0
∗
+ O2,0
∗
ζ 1 O1,0
∗
+ O2,0
∗
d o y o o
P1,t−1 Y1,t − P1,t−1 O1,t + P1,t−1 Y1,t
GDP1,t = GDP1,t−1 d o y o o
. (90)
P1,t−1 Y1,t−1 − P1,t−1 O1,t−1 + P1,t−1 Y1,t−1
13
The linearized equation is given by:
o∗ y∗ o∗ o∗ µ
P1,0 O1,0 P1,0 Y1,0
1 − d∗ d∗ + d∗ d∗ d d
gdp1,t − gdp1,t−1 − z
ŷ1,t − ŷ1,t−1
P1,0 Y1,0 P1,0 Y1,0 µgdp,1
o∗ y∗ o∗ o∗
P1,0 O1,0 µo y y P1,0 Y1,0 µo o o
= − d∗ d∗ ô − ôt−1 + d∗ d∗ ŷ − ŷ1,t−1
P1,0 Y1,0 µgdp,1 t P1,0 Y1,0 µgdp,1 1,t
o∗ y∗ o∗ o∗ \
P1,0 O1,0 P1,0 Y1,0 µo P1o
− d∗ d∗
− d∗ d∗ −1 . (91)
P1,0 Y1,0 P1,0 Y1,0 µgdp,1 P1d t−1
Absent trend growth the linear approximation of the Laspeyres index for GDP is a constant price
aggregate.
∗ \ o∗ y∗ o∗ o∗ \ !
N GDP1,0 N GDP1 P1,0 O1,0 P1,0 Y1,0 P1o
d∗ d∗
= − d∗ d∗ ŷ1,t − (93)
P1,0 Y1,0 P1d Y1 t d∗ d∗
P1,0 Y1,0 P1,0 Y1,0 P1d t
o∗ y∗ o∗ o∗
P1,0 O1,0 y P1,0 Y1,0 o
− ô
d∗ d∗ 1,t
+ d∗ d∗ 1,t
ŷ . (94)
P1,0 Y1,0 P1,0 Y1,0
14
43. Trade balance to GDP ratio:
bal bal d
N T1,t N T1,t P1,t Y1,t
= d . (97)
N GDP1,t P1t Y1,t N GDP1,t
obs
log GDP1,t obs
− log GDP1,t−1 d 1,t − gdp
= gdp d 1,t−1 + µgdp,1 − 1 . (101)
15
The linearized equation is given by:
" #
\ d∗ d∗ ∗ o∗ \
P1o (O o
1 − Y1 ) P1,0 Y1,0 O1,0 Y1,0 N\
GDP1 P1o
= d∗
− ∗ − + − ŷ1,t
N GDP1 t N GDP1,0 ∗
Y1,0 Y1,0 P1d Y1 t P1d t
d∗ ∗ ∗ o∗
P1,0 Y1,0 O1,0 Y1,0 o
+ ô − ∗ ŷ1,t .
d∗ 1t
(104)
N GDP1,0 ∗
Y1,0 Y1,0
d M
P1,t 1,t
49. Observation of nonoil import share N GDP1,t
:
d d d
P1,t M1,t P1,t M1,t P1,t Y1,t
= d . (106)
N GDP1,t P1,t Y1,t N GDP1,t
d X
P1,t 1,t
50. Observation of nonoil export share N GDP1,t
:
d d d
P1,t X1,t P1,t X1,t P1,t Y1,t
= d . (108)
N GDP1,t P1,t Y1,t N GDP1,t
c obs
rer c 1,t .
1,t = rer (110)
16
c C
P1,t 1,t
52. Observation of consumption share N GDP1,t
:
c c d
P1,t C1,t P1,t C1,t P1,t Y1,t
= d . (111)
N GDP1,t P1,t Y1,t N GDP1,t
i I
P1,t 1,t
53. Observation of (fixed) investment share in GDP N GDP1,t
:
i c d
P1,t I1,t P1,t I1,t P1,t Y1,t
= d . (113)
N GDP1,t P1,t Y1,t N GDP1,t
π core,obs
1,t = π core
1,t + (π 1 − 1).
∗
(115)
ω obs
1,t = ω 1,t . (116)
s,obs s
r1,t = r1,t . (117)
17
A.1.5 Decomposition of Marginal Costs
ŵ1,t = mc d 1,t .
c 1,t + mpl (125)
since ω oy vy vy
1 + ω 1 φ1 + ω 1 (1 − φ1 ) = 1.
Along the balanced growth path real quantities grow at the common rate µz , except for oil demand
and supply, and hours worked. Prices (relative to the domestic good), including real marginal costs, are
constant except for real wages and the real price of oil. With labor augmenting technological progress,
hours worked are stationary and real wages need to grow with the common growth rate µz . Oil supply
µz
and oil demand grow at the rate µo < µz , while the price of oil grows at the rate µzo < µo
. Nominal
prices grow at the inflation rate π ∗1 . Below, we define relationships that need to hold along the balanced
growth path. Unless noted otherwise, an expression presented for the home country s identical to the
one for the foreign country. The size of country 1 relative to that of country 2 is denoted by ζ 1 .
Xi,t
∗
refers to the value of variable Xi at time t along the balanced growth path. Xi,0
∗
is the initial
value of the variable at time 0.
18
A.2.1 Calibrated Expressions
Some parameters in our model are not estimated, but are implicitly pinned down by assigning data means
to the following expressions.
9. Normalization of ω l1 :
ω l1 = 1. (135)
19
A.2.2 Composite Parameters
Given the parameter choices and the expressions described above, the remaining parameters of the model
can be computed as shown below.
y∗ y∗
µtzo O1,t O1,0
ω oy
1= d∗
= d∗ = shareoyy1 (136)
Y1,t Y1,0
1
shareoyy1 = shareoy1 1 − . (137)
1 + ratiooyoc1
V1,t
∗
V1,0
∗
ω vy oy
1 = 1 − ω1 = = = sharevy1 . (138)
Y1,t
∗d
Y1,0
∗d
k∗
3. From condition 6 compute r1,0 :
k∗ µz
r1,0 = − 1 + δ1. (139)
β1
20
8. From condition 12 define shareocc1 and compute ω oc
1 :
µtzo O1,t
c∗ c∗
O1,0
ω oc
1 = = = shareocc1 , (145)
C1,t
∗
C1,0
∗
where
shareoy1 − shareoyy1
shareocc1 = . (146)
sharecy1
ne∗ ne∗
C1,t C1,0
ω cc
1 =1− ω oc
1 = ∗ = ∗ . (147)
C1,t C1,0
sharemy1 1
sharemccn1 = . (150)
sharecy1 sharecnc1 1 + ratiomimc1
where
sharemy1 ratiomimc1
sharemii1 = . (153)
shareiy1 1 + ratiomimc1
21
14. Define exports relative of gross output country 1 sharexy1 :
d∗ ∗ d∗ o∗ ∗ o∗ o∗
P1,t X1,t P1,t M1,t
∗
P1,t O1,t P1,t Y1,t
d∗ d
= d∗ d∗
+ d∗ d∗
− d∗ d∗
= sharexy1 , (154)
P1,t Y1,t P1,t Y1,t P1,t Y1,t P1,t Y1,t
or
or
d∗
Y1,0
sharexy2 = sharemy2 − (shareoy1 − shareyoy1 ) d∗ ζ 1 . (157)
Y2,0
16. If trade is balanced along the balanced growth path, we define sharemy2 :
d∗ o∗ o∗
d∗ d∗ d∗
P1,t M1,t
∗
P1,t O1,t
∗
− Y1,t e∗1,t P2,t M2,t
∗
P2,t Y2,t 1
d∗ d∗
+ d∗ d∗
= d∗ d∗ d∗ d∗ ζ
= sharemy2 , (158)
P1,t Y1,t P1,t Y1,t P2,t Y2,t P1,t Y1,t 1
or
d∗
Y1,0
sharemy2 = (sharemy1 + (shareoy1 − shareyoy1 )) d∗ ζ 1 (159)
Y2,0
or
µz − (shareoyy1 − shareyoy1 ) µo
µgdp,1 = . (162)
1 − (shareoyy1 − shareyoy1 )
22
d∗
Y1,t
19. Gross output ratio d∗
Y2,t
:
d∗ d∗
Y1,t Y1,0 sharely2 L∗1,0
d∗
= d∗
= . (163)
Y2,t Y2,0 sharely1 L∗2,0
20. From condition 39 define the share in world oil production for country 2 shareoprod2 :
o∗
Y2,0
o∗
Y2,t Y d∗
o∗ o∗
= Y o∗ Y 2,0d∗ o∗ = shareoprod2 , (164)
ζ 1 Y1,t + Y2,t ζ 1 Y1,0
Y2,0
d∗ Y d∗ +
1,0
d∗
Y2,0
1,0 2,0
or
shareyoy2
shareoprod2 = Y d∗
. (165)
shareyoy2 + shareyoy1 ζ 1 Y1,0
d∗
2,0
21. From condition 39 define the share in world oil consumption for country 2 shareocon2 :
O2,0
∗
O2,t∗ d∗
Y2,0
o∗
= d∗ = shareocon2 , (166)
ζ 1 O1,t
∗
+ Y2,t O ∗
ζ 1 Y 1,0
Y1,0 O2,0
∗
d∗ Y d∗ + d∗
Y2,0
1,0 2,0
or
shareoy2
shareocon2 = Y d∗
. (167)
shareoy2 + shareoy1 ζ 1 Y1,0
d∗
2,0
d∗
= d∗
= shareyoy1 , (168)
Y1,t O1,t
∗
Y1,t
or
23
B Appendix: Data
The model is estimated by the method of maximum likelihood. The data are quarterly and run between
1984 and the third quarter of 2008. A presample of 10 years from 1974 to 1984 is used to train the
Kalman filter used to form the likelihood. The observed series are the following:
1. the log of U.S. real GDP, from NIPA Table 1.1.3 (line 1);
2. the log of trade-weighted foreign GDP. The series reflects GDP data from national sources for the
26 most important trading partners of the United States. A description of the export weights is
in Loretan (2005). The countries included account for well over 90% of U.S. exports, as well as
imports.
3. the log of the U.S. real dollar price of oil defined as the refiners’ acquisition cost for imported crude
from the U.S. Energy Information Administration
(http : //www.eia.doe.gov/dnav/pet/pet pri rac2 dcu nus m.htm) normalized by the GDP defla-
tor from NIPA Table 1.1.4 (line 1);
4. the log of U.S. crude oil production from Table 11.1b of the Monthly Energy Review of the U.S.
Energy Information Administration
(http : //www.eia.doe.gov/totalenergy/data/monthly/#petroleum);
5. the log of foreign oil production (calculated as world production net of U.S. production) from
Table 11.b of the Monthly Energy Review of the U.S. Energy Information Administration (http :
//www.eia.doe.gov/totalenergy/data/monthly/#petroleum);
6. the log of U.S. hours worked in the nonfarm business sector from the Labor Productivity and Cost
Database of the U.S. Bureau of Labor Statistics, normalized by the U.S. civilian non-institutional
population from the U.S. Bureau of Labor Statistics
(http : //data.bls.gov/pdq/querytool.jsp?survey = pr);
7. the log of the broad real dollar exchange rate from the U.S. Federal Reserve Board
(http : //www.f ederalreserve.gov/releases/H10/summary/, the weights are the same as the ones
24
described for the measure of foreign GDP above);
8. U.S. personal consumption expenditures from NIPA Table 1.1.5 (line 2), expressed as a share of
U.S GDP from NIPA Table 1.1.5 (line 1);
10. U.S. imports of non-petroleum goods from NIPA Table 4.2.5 (line 54), expressed as a share of GDP;
11. U.S. goods exports from NIPA Table 4.2.5 (line 2), expressed as a share of GDP;
12. U.S. fixed investment from NIPA Table 4.2.5 (line 8), expressed as a share of GDP;
13. U.S. core inflation measured as the log change in the deflator for personal consumption expenditures
excluding food and energy prices from the Federal Reserve Bank of St. Louis Fred Database,
14. U.S. wage inflation (demeaned) measured using the log change in nominal compensation from the
Labor Productivity and Cost Database of the U.S. Bureau of Labor Statistics
(http : //data.bls.gov/pdq/querytool.jsp?survey = pr);
15. and the U.S. effective federal funds rate (demeaned) from the Federal Reserve Board (http :
//www.f ederalreserve.gov/releases/h15/data.htm).
25
C Appendix: Plots of Data and In-Sample Forecast Errors
This appendix contains plots of all the observed data, the in-sample 1 step ahead forecasts, and the
forecast errors.
26
Figure 1: Data and Forecast Errors
U.S. GDP
0.8
Data
0.7 Fitted Values (in sample, 1 step ahead forecast)
0.6
0.5
Log Scale
0.4
0.3
0.2
0.1
0
1985 1990 1995 2000 2005
Forecast Error
0.015
0.01
0.005
Log Scale
−0.005
−0.01
−0.015
−0.02
1985 1990 1995 2000 2005
27
Figure 2: Data and Forecast Errors
Foreign GDP
0.9
Data
0.8 Fitted Values (in sample, 1 step ahead forecast)
0.7
0.6
0.5
Log Scale
0.4
0.3
0.2
0.1
−0.1
1985 1990 1995 2000 2005
Forecast Error
0.02
0.015
0.01
Log Scale
0.005
−0.005
−0.01
1985 1990 1995 2000 2005
28
Figure 3: Data and Forecast Errors
U.S. Oil Production
0.1
Data
0 Fitted Values (in sample, 1 step ahead forecast)
−0.1
−0.2
Log Scale
−0.3
−0.4
−0.5
−0.6
−0.7
1985 1990 1995 2000 2005
Forecast Error
0.06
0.04
0.02
−0.02
Log Scale
−0.04
−0.06
−0.08
−0.1
−0.12
−0.14
1985 1990 1995 2000 2005
29
Figure 4: Data and Forecast Errors
Foreign Oil Production
0.45
Data
0.4 Fitted Values (in sample, 1 step ahead forecast)
0.35
0.3
0.25
Log Scale
0.2
0.15
0.1
0.05
−0.05
1985 1990 1995 2000 2005
Forecast Error
0.08
0.06
0.04
Log Scale
0.02
−0.02
−0.04
−0.06
1985 1990 1995 2000 2005
30
Figure 5: Data and Forecast Errors
U.S. Oil Imports (GDP share)
3
Data
Fitted Values (in sample, 1 step ahead forecast)
2.5
2
Percent
1.5
0.5
0
1985 1990 1995 2000 2005
Forecast Error
0.8
0.6
0.4
0.2
Percent
−0.2
−0.4
−0.6
1985 1990 1995 2000 2005
31
Figure 6: Data and Forecast Errors
Real Oil Price
1
Data
Fitted Values (in sample, 1 step ahead forecast)
0.5
0
Log Scale
−0.5
−1
−1.5
1985 1990 1995 2000 2005
Forecast Error
0.4
0.3
0.2
0.1
Log Scale
−0.1
−0.2
−0.3
−0.4
−0.5
1985 1990 1995 2000 2005
32
Figure 7: Data and Forecast Errors
U.S. Nonoil Goods Imports (GDP share)
13
Data
Fitted Values (in sample, 1 step ahead forecast)
12
11
10
Percent
6
1985 1990 1995 2000 2005
Forecast Error
0.6
0.4
0.2
0
Percent
−0.2
−0.4
−0.6
−0.8
1985 1990 1995 2000 2005
33
Figure 8: Data and Forecast Errors
U.S. Goods Exports (GDP share)
10
Data
Fitted Values (in sample, 1 step ahead forecast)
9
8
Percent
4
1985 1990 1995 2000 2005
Forecast Error
0.6
0.4
0.2
0
Percent
−0.2
−0.4
−0.6
−0.8
1985 1990 1995 2000 2005
34
Figure 9: Data and Forecast Errors
U.S. Real Exchange Rate
0.3
Data
0.25 Fitted Values (in sample, 1 step ahead forecast)
0.2
0.15
Log Scale
0.1
0.05
−0.05
−0.1
−0.15
1985 1990 1995 2000 2005
Forecast Error
0.05
0.04
0.03
0.02
0.01
Log Scale
−0.01
−0.02
−0.03
−0.04
−0.05
1985 1990 1995 2000 2005
35
Figure 10: Data and Forecast Errors
U.S. Private Consumption Expenditures (GDP share)
72
Data
71 Fitted Values (in sample, 1 step ahead forecast)
70
69
Percent
68
67
66
65
64
63
1985 1990 1995 2000 2005
Forecast Error
1
0.5
0
Percent
−0.5
−1
−1.5
1985 1990 1995 2000 2005
36
Figure 11: Data and Forecast Errors
U.S. Fixed Investment (GDP share)
18
Data
Fitted Values (in sample, 1 step ahead forecast)
17
16
Percent
15
14
13
12
1985 1990 1995 2000 2005
Forecast Error
0.6
0.5
0.4
0.3
0.2
Percent
0.1
−0.1
−0.2
−0.3
−0.4
1985 1990 1995 2000 2005
37
Figure 12: Data and Forecast Errors
U.S. Core Inflation (qr)
2
Data
Fitted Values (in sample, 1 step ahead forecast)
1.5
1
Percent
0.5
−0.5
1985 1990 1995 2000 2005
Forecast Error
0.8
0.6
0.4
0.2
Percent
−0.2
−0.4
−0.6
−0.8
−1
1985 1990 1995 2000 2005
38
Figure 13: Data and Forecast Errors
U.S. Federal Funds Rate (qr)
3
Data
Fitted Values (in sample, 1 step ahead forecast)
2.5
2
Percent
1.5
0.5
0
1985 1990 1995 2000 2005
Forecast Error
0.3
0.2
0.1
0
Percent
−0.1
−0.2
−0.3
−0.4
−0.5
1985 1990 1995 2000 2005
39
Figure 14: Data and Forecast Errors
U.S. Wage Inflation (qr)
4
Data
3.5 Fitted Values (in sample, 1 step ahead forecast)
2.5
Percent
1.5
0.5
−0.5
1985 1990 1995 2000 2005
Forecast Error
2
1.5
0.5
0
Percent
−0.5
−1
−1.5
−2
−2.5
−3
1985 1990 1995 2000 2005
40
Figure 15: Data and Forecast Errors
U.S. Hours Worked (PC)
0.14
Data
0.12 Fitted Values (in sample, 1 step ahead forecast)
0.1
0.08
Log Scale
0.06
0.04
0.02
−0.02
1985 1990 1995 2000 2005
Forecast Error
0.01
0.005
0
Log Scale
−0.005
−0.01
−0.015
1985 1990 1995 2000 2005
41
D Appendix: Plots of Smoothed Estimates of Shock Processes
and Their Innovations
This appendix contains plots of all the shock processes and their innovations. A key to the symbols
denoting each shock process is given in Table 1 in the main body of the paper.
42
Figure 16: Data and Forecast Errors
Shock Process Z1,t
0.22
0.2
0.18
0.16
0.14
Log Scale
0.12
0.1
0.08
0.06
0.04
0.02
1985 1990 1995 2000 2005
1.5
1
Log Scale
0.5
−0.5
−1
−1.5
−2
1985 1990 1995 2000 2005
43
Figure 17: Data and Forecast Errors
Shock Process Z2,t
0.45
0.4
0.35
0.3
Log Scale
0.25
0.2
0.15
0.1
0.05
1985 1990 1995 2000 2005
1
Log Scale
−1
−2
−3
1985 1990 1995 2000 2005
44
Figure 18: Data and Forecast Errors
o
Shock Process Y1,t
0
−0.2
−0.4
Log Scale
−0.6
−0.8
−1
−1.2
−1.4
1985 1990 1995 2000 2005
o
Innovations to Shock Process Y1,t
3
0
Log Scale
−1
−2
−3
−4
−5
1985 1990 1995 2000 2005
45
Figure 19: Data and Forecast Errors
o
Shock Process Y2,t
0.1
0.05
0
Log Scale
−0.05
−0.1
−0.15
−0.2
−0.25
1985 1990 1995 2000 2005
o
Innovations to Shock Process Y2,t
4
2
Log Scale
−1
−2
−3
1985 1990 1995 2000 2005
46
Figure 20: Data and Forecast Errors
o
Shock Process Z1,t
1.05
0.95
0.9
0.85
Log Scale
0.8
0.75
0.7
0.65
0.6
0.55
1985 1990 1995 2000 2005
o
Innovations to Shock Process Z1,t
3
1
Log Scale
−1
−2
−3
−4
1985 1990 1995 2000 2005
47
Figure 21: Data and Forecast Errors
o
Shock Process Z2,t
2
1.8
1.6
1.4
1.2
Log Scale
0.8
0.6
0.4
0.2
0
1985 1990 1995 2000 2005
o
Innovations to Shock Process Z2,t
4
2
Log Scale
−1
−2
1985 1990 1995 2000 2005
48
Figure 22: Data and Forecast Errors
c
Shock Process Z1,t
0.05
0.045
0.04
0.035
0.03
Log Scale
0.025
0.02
0.015
0.01
0.005
0
1985 1990 1995 2000 2005
c
Innovations to Shock Process Z1,t
4
2
Log Scale
−1
−2
1985 1990 1995 2000 2005
49
Figure 23: Data and Forecast Errors
c
Shock Process Z2,t
0.03
0.02
0.01
0
Log Scale
−0.01
−0.02
−0.03
−0.04
−0.05
−0.06
1985 1990 1995 2000 2005
c
Innovations to Shock Process Z2,t
5
2
Log Scale
−1
−2
−3
1985 1990 1995 2000 2005
50
Figure 24: Data and Forecast Errors
m
Shock Process Z1,t
0
−0.1
−0.2
Log Scale
−0.3
−0.4
−0.5
−0.6
−0.7
1985 1990 1995 2000 2005
m
Innovations to Shock Process Z1,t
3
1
Log Scale
−1
−2
−3
1985 1990 1995 2000 2005
51
Figure 25: Data and Forecast Errors
m
Shock Process Z2,t
0.3
0.2
0.1
0
Log Scale
−0.1
−0.2
−0.3
−0.4
−0.5
1985 1990 1995 2000 2005
m
Innovations to Shock Process Z2,t
2
1.5
0.5
Log Scale
−0.5
−1
−1.5
−2
−2.5
1985 1990 1995 2000 2005
52
Figure 26: Data and Forecast Errors
w
Shock Process θ1,t
20
10
0
Log Scale
−10
−20
−30
−40
1985 1990 1995 2000 2005
w
Innovations to Shock Process θ1,t
3
1
Log Scale
−1
−2
−3
1985 1990 1995 2000 2005
53
Figure 27: Data and Forecast Errors
p
Shock Process θ1,t
1.5
0.5
Log Scale
−0.5
−1
−1.5
1985 1990 1995 2000 2005
p
Innovations to Shock Process θ1,t
3
2.5
1.5
1
Log Scale
0.5
−0.5
−1
−1.5
−2
1985 1990 1995 2000 2005
54
Figure 28: Data and Forecast Errors
Shock Process π̄1,t
0.06
0.04
0.02
0
Log Scale
−0.02
−0.04
−0.06
−0.08
−0.1
1985 1990 1995 2000 2005
1
Log Scale
−1
−2
−3
−4
1985 1990 1995 2000 2005
55
Figure 29: Data and Forecast Errors
i
Shock Process Z1,t
0.3
0.25
0.2
Log Scale
0.15
0.1
0.05
−0.05
1985 1990 1995 2000 2005
i
Innovations to Shock Process Z1,t
4
2
Log Scale
−1
−2
1985 1990 1995 2000 2005
56
Figure 30: Data and Forecast Errors
g
Shock Process Z1,t
0.2
0.15
0.1
0.05
Log Scale
−0.05
−0.1
−0.15
−0.2
1985 1990 1995 2000 2005
g
Innovations to Shock Process Z1,t
3
1
Log Scale
−1
−2
−3
1985 1990 1995 2000 2005
57
E Appendix: Interpreting Oil Efficiency Shocks
While oil efficiency shocks are meant to capture fundamental changes in the demand for oil – such as
a shift towards motorization in emerging Asia, continuing industrialization of China, or energy-efficient
cars becoming more popular in the United States – we would like to exclude the possibility that the
efficiency shocks are nothing but a catchall for all unmodeled disturbances that can affect oil demand.
Accordingly, we conduct a simple exogeneity test for our estimated oil efficiency innovations εzo
1,t
and εzo
2,t following the exercise in Evans (1992) for technology shocks. Let
εzo zo
i,t = Ai (L)εi,t−1 + Bi (L)xt−1 + ν i,t , (170)
where ν i,t is a mean zero, i.i.d random variable, Ai (L) and Bi (L) are polynomials in the lag operator L,
x is a vector that includes potential explanatory variables for oil demand. Ideally, none of the variables
included in xt−1 should Granger cause the innovations εzo zo
1,t and ε2,t .
Unmodeled determinants of oil demand that we would like to see assessed as unimportant for our
measure of oil efficiency are: (i) log-changes in oil inventories (OECD and U.S. inventory data), (ii) log-
changes of the price of oil substitutes in energy generation (price of coal), (iii) financial market activities
(log-changes in oil futures, real stock market returns), (iv) changes and the level of activity-driven demand
for all commodities (index proposed in Kilian (2009)) as these should already be captured by our GDP
measures.1
When including each of the variables separately with up to three lags, the null hypothesis of
exogeneity, B2 (L) = 0, was never rejected for the foreign oil efficiency shocks at the 5% significance
level.2 For the U.S. oil efficiency shock, the null hypothesis was rejected at the 5% level of significance for
1
The variables in xt−1 in equation (170) enter either as a return or in growth rates. The data sources are as follows.
Oil inventory: U.S. Energy Information Administration; given the lack of data on crude oil inventories for countries other
than the United States, we construct OECD crude oil inventories by scaling OECD petroleum stocks over U.S. petroleum
stocks for each period as suggested in Hamilton (2009), and Kilian and Murphy (2010). Price of coal: Australian Coal,
World Bank Commodity Price Data (Pink Sheet). Oil futures: three months NYMAX contract, U.S. Energy Information
Administration. Stock market index: S&P 500, FRED database St. Louis Fed. World commodity demand: index of global
real economic activity in industrial commodity markets, available at http://www-personal.umich.edu/ lkilian/reaupdate.txt.
2
A detailed description of Granger causality tests in a multivariate context is given in Hamilton (1994), chapter 11.
Let σ̂ ν i (0) be the estimate of the variance of the innovations, σ ν i under the null hypothesis and σ̂ ν i be the unrestricted
estimate. The test statistic T {ln|σ̂ ν i (0)| − ln|σ̂ ν i |} follows a χ2 distribution with the degrees of freedom, df, given by the
product of the number of lags and the number of variables included in x. If the test statistic exceeds the α% critical value
for χ2 (df ), the block exogeneity assumption is rejected at the α% level of significance.
58
log-changes in OECD oil inventories and changes in Kilian’s shipping index. Yet, the adjusted R2 never
exceeds 11%. With up to three lags in the VAR, no combination of variables ever implied a rejection
of the null hypothesis at conventional significance levels for the foreign oil efficiency shock (up to 10%).
Moreover, the adjusted R2 never exceeds 12%. For the U.S. oil efficiency shock, the most challenging
combination of variables is given by the set of changes in Kilian’s activity measure, and log-changes
in OECD and U.S. oil inventories. In this case, the null hypothesis of block exogeneity is rejected at
the 1% significance level for each lag-length up to three lags. However, with adjusted R2 values not
exceeding 15%, the combined explanatory power of these variables is low. We conclude that none of the
combinations we tested poses an important challenge to our interpretation of the oil efficiency shocks.3
3
Further increases in the lag-length not only did not overturn our overall assessment, but reinforced it, instead.
59
References
Evans, C. L. (1992). Productivity Shocks and Real Business Cycles. Journal of Monetary Economics 29,
191–208.
Hamilton, J. (1994). Time Series Analysis. Princeton, NJ: Princeton University Press.
Hamilton, J. (2009). Causes and Consequences of the Oil Shock of 2007-08. Brookings Papers on
Economic Activity (1), 215–84.
Kilian, L. (2009). Not All Price Shocks Are Alike: Disentangling Demand and Supply Shocks in the
Crude Oil Market. American Economic Review 99 (3), 1053–69.
Kilian, L. and D. Murphy (2010). The Role of Inventories and Speculative Trading in the Global Market
for Crude Oil. Mimeo, University of Michigan.
Loretan, M. (2005). Indexes of the Foreign Echange Value of the Dollar. Federal Reserve Bulletin, 1–8.
Smets, F. and R. Wouters (2007). Shocks and Frictions in US Business Cycles: A Bayesian DSGE
Approach. American Economic Review 97 (3), 586–606.
60