Solution techniquesMartin Ellison
University of Warwick and CEPR
Bank of England, December 2005
State-space form
10110 tttt vBXAXEA
Generalised state-space form
Many techniques available to solve this class of models
We use industry standard: Blanchard-Kahn
Alternative state-space form
10101
101
tttt vBAXAAXE
11 tttt BvAXXE
A B
Partitioning of model
11
1
t
t
t
tt
t Bvyw
AyEw
t
tt y
wX
backward-looking variables
predetermined variables
forward-looking variablescontrol variables
Jordan decomposition of A
11
1
t
t
t
tt
t Bvyw
AyEw
1 PPA
eigenvectors diagonal matrix of eigenvalues
Blanchard-Kahn condition
The solution of the rational expectations model is unique if the number of unstable eigenvectors of the system is exactly equal to the number of
forward-looking (control) variables.
i.e., number of eigenvalues in Λ greater than 1 in magnitude must be equal to number of forward-looking variables
tw
ty
Too many stable roots
0w
multiple solutions
equilibrium path not unique
need alternative techniques
tw
ty
Too many unstable roots
0w
no solution
all paths are explosive
transversality conditions violated
tw
ty
Blanchard-Kahn satisfied
0w
one solution
equilibrium path is unique
system has saddle path stability
Rearrangement of Jordan form
111
1
11
t
t
t
tt
t BvPyw
PyEw
P
11
1
1
t
t
t
tt
t Bvyw
PPyEw
R
Partition of model
2
1
00
*22
*21
*12
*111
PPPP
P
11
1
11
t
t
t
tt
t Rvyw
PyEw
P
2
1
RR
R
stable
unstable
Transformed problem
1
1~~
t
tt yw
E
t
t
yw~~
ttt
ttt
yyPwP
wyPwP~
~*22
*21
*12
*11
12
1*22
*21
*12
*11
2
1
1
1*22
*21
*12
*11
00
t
t
t
tt
t vRR
yw
PPPP
yEw
PPPP
12
1
2
1
1
1~~
00
~~
t
t
t
tt
t vRR
yw
yEw
Decoupled equations
12
1
2
1
1
1~~
00
~~
t
t
t
tt
t vRR
yw
yEw
1111~~
ttt vRww
1221~~
tttt vRyyE
Decoupled equations can be solved separately
stable
unstable
Solution strategy
Solve unstable transformed equation ty~
Translate back into original problem
tw~
t
t
yw
Solve stable transformed equation
Solution of unstable equation
As , only stable solution is12 tyt 0~
0~ *22
*21 ttt yPwPy
tj
jtt yyE ~~2
tt wPPy *21
1*22
Solve unstable equation forward to time t+j
Forward-looking (control) variables are function of backward-looking (predetermined) variables
Solution of stable equation
tt
ttt
wPPy
yPwPw*21
1*22
*12
*11
~
tj
jtt wwE ~~1
tt wPPPPw )(~ *21
1*22
*12
*11
Solve stable equation forward to time t+j
As , no problems with instability11
Solution of stable equation
111*
211*
22*12
*11
*21
1*22
*12
*111
1*21
1*22
*12
*111
)(
)()(
t
tt
vRPPPP
wPPPPPPPPw
1111~~
ttt vRww
tt wPPPPw )(~ *21
1*22
*12
*11
11*21
1*22
*12
*11
~)( tt wwPPPP
Future backward-looking (predetermined) variables are function of current backward-looking (predetermined) variables
Full solution
111*
211*
22*12
*11
*21
1*22
*12
*111
1*21
1*22
*12
*111
*21
1*22
)(
)()(
t
tt
tt
vRPPPP
wPPPPPPPPw
wPPy
All variables are function of backward-looking (predetermined) variables: recursive structure
Baseline DSGE model
tt
t
tt
tt vx
ExE
0ˆˆ
11
ˆˆ
01 11
1
11
11 ttt vv
State space form
To make model more interesting, assume policy shocks vt follow an AR(1) process
New state-space form
tv
111
1
1
11
001
ˆˆ
101
00
ˆˆ
0010
001
t
t
t
t
tt
tt
t
xv
ExEv
ttx ,ˆ
One backward-looking variable
Two forward-looking variables
Blanchard-Khan conditions
t
tt
tt
xy
vw
ˆ
Require one stable root and two unstable roots
Partition model according to
Next steps
Exercise to check Blanchard-Kahn conditions numerically in MATLAB
Numerical solution of model
Simulation techniques