Professional Documents
Culture Documents
Anton Izvolenskiy
Andres Fernando Garcia Parrado
Dimitrios Bermperoglou
1. A short theoretical background of short run effects of money
money and
interest rate on economy
Instruments : variables directly controlled by the central bank (e.g. an interest rate
charged on reserves borrowed by the central bank)
We are going to use both money supply and money market interest rate as the policy
variables.
Open-market operations by the CB are the principal tool of monetary policy: the CB can
increase or reduce the monetary base by buying government debt from banks or selling
government debt to banks.
1
Monetary policy and its short-
short-run effects
If the ECB pursues expansionary monetary policy by increasing the supply of money
then the nominal interest rate will fall, investment will rise, consumption will rise. Finally,
output may also rise (see IS-LM model).
In long-run, prices increase and restore equilibrium. Money does not affect real
economy.
2
Contractionary Monetary Policy
As the money supply is contracted, interest rates rise, investment will fall, consumption
will fall.
In turn…rising
turn…rising interest
interest rates
People will choose to hold more bonds than they did before because the return on a
bond has risen (i has gone up).
So, if the ECB pursues contractionary monetary policy by decreasing the supp
supply
ly of
money, the nominal interest rate will rise, investment will fall, and consumption will fall
and so output.
output
3
2. Research
Sims (1992)
Eichenbaum (1992)
4
3. Econometric project
Data sources
International Monetary Fund
3 countries: Estonia, Latvia, Lithuania
Quarterly data from 1998 – 2009
Two samples a) before EU accession 1998:1 – 2004:2
b) after EU 2004:3 – 2009:4
Variables: Money supply M1 (Currency in circulation + overnight deposits)
Annual rate of change of HCPI
Money market interest rate
Per capita GDP in current prices 2005
Note: The Eastern-european countries do not have sufficient data available since
their economic record (history) is too short till now. For this reason, the results we
obtain are not very robust. However, we do make some comments on them, just to
emphasize how empirical work (impulse responses here) can be linked and directly
compared with theoretical facts.
5
Evidence for the causality between money growth and output
.20
.15
.10
.05
.00
-.05
-.10
-.15
-.20
98 99 00 01 02 03 04 05 06 07 08 09
LM GDP
6
Methodology
Yt = µ + δt + Π1 ⋅ Yt −1 + Π 2 ⋅ Yt − 2 + Π 3 ⋅ Yt − 3 + Π 4 ⋅ Yt − 4 + ut
Where
mt
Yi - vector of variables π t
yt
• First we have to decide the optimal lag length of the VAR model. Doing it through E-views, we
obtain the following table. As we see, the most information criteria suggest using 4 lags.
7
• Next, we estimate the VAR(4) model and the coefficients estimates are summarized in a more
convenient VAR representation as follows:
mt − 0.39 0.05 0.36 0.13 0.06 mt −1 0.14 0.05 0.009 mt −2
π = 0.32 + 5 E − 05 ⋅ t + − 0.11 0.51 0.07 ⋅ π + 0.13 − 0.25 − 0.03 ⋅ π +
t t −1 t −2
yt 0.06 0.13 0.39 1.56 − 0.24 y t −1 − 0.37 − 0.63 0.48 y t −2
− 0.22 1.93 0.05 mt −3 0.11 − 2.03 0.56 mt −4 u mt
+ − 0.03 0.89 − 0.01 ⋅ π t −3 + 0.02 − 0.83 − 0.08 ⋅ π t −4 + u πt
− 0.22 − 0.17 0.14 y t −3 0.38 − 0.06 0.42 y t − 4 u yt
σ m2 σ mπ σ my
Vu = σ mπ σ π2 σ πy
σ my σ πy σ y2
• Remember that VAR presented above can be reduced to a vector notation form as:
Yt = µ + δt + Π1 ⋅ Yt −1 + Π 2 ⋅ Yt − 2 + Π 3 ⋅ Yt − 3 + Π 4 ⋅ Yt − 4 + ut
We define a lag operator L such that: Lxt = xt −1
Hence, the above matrix equation becomes:
Yt = µ + δt + Π1 ⋅ Yt −1 + Π 2 ⋅ L Yt −1 + Π 3 ⋅ L2Yt −1 + Π 4 ⋅ L3Yt −1 + ut
(
⇔ Yt = µ + δt + Π1 + Π 2 L + Π 3 L2 + Π 4 L3 ⋅ Yt −1 + ut )
⇔ Yt = µ + δt + Π ( L) ⋅ Yt −1 + ut
• However, in order to compute impulse responses, we need to estimate the coefficients of the
SMA(∞) representation. Structural moving average (SMA) representation is derived through the
SVAR(p) representation. And SVAR(p), in turn, can be derived from the VAR(p).
A SVAR representation can be derived from the reduced VAR by multiplying the VAR model
with a Q matrix.
8
Q ⋅ Yt = Q ⋅ µ + Q ⋅ δt + Q ⋅ Π ( L) ⋅ Yt −1 + Q ⋅ ut
= κ + т + Q ⋅ Π ( L) ⋅ Yt −1 + et
σ~m2 0 0
~
Ve = 0 σ2
π 0
0 0 σ~ y2
As we see, the structural VAR captures not only a lagged but also contemporaneous
relationships among all the variables of the model. The contemporaneous effects are
summarized in the matrix Q. Also the VAR shocks are related to the structural shocks with the
formula:
Qu = e
Obviously, Q is a 3x3 matrix since there are 3 variables in the model.
Now, we have to make some identification assumptions on matrix Q, in order to exactly
identify the SVAR parameters using the estimated VAR parameters. For this reason, we proceed
to the following manipulations:
Qu = e ⇔ u = Q −1e = Θ ⋅ e
Identification Assumption:
Assumption Monetary policy is exogenous to contemporaneous output
and inflation shocks.
Θ is lower triangular
u m θ11 0 0 ε m u m = θ11ε m
uπ = θ 21 θ 22 0 ⋅ ε π ⇒ u π = θ 21ε m + θ 22 ε π
u y θ 31 θ 32 θ 33 ε y u = θ ε + θ ε + θ ε
y 31 m 32 π 33 y
9
According to our assumption of the exogeneity of the policy shock, we need to make Θ (thus Q)
lower triangular. In practice, that means setting 3 elements of the matrix equal to zero.
However, E-views uses a slightly different notation of presenting the relation between the VAR
and the SVAR shocks. It assumes that:
A⋅u = B ⋅e
So nothing changes in essence. What was previously as matrix Q now can be decomposed as:
Q = B −1 A
Estimated A matrix:
1.000000 0.000000 0.000000
0.197561 1.000000 0.000000
-1.404929 -1.994283 1.000000
Estimated B matrix:
0.017921 0.000000 0.000000
0.000000 0.009269 0.000000
0.000000 0.000000 0.014260
10
Note that with the help of matrices A and B (thus Q) the VAR can be written as
Y t = µ + δ t + Π ( L ) ⋅ Y t −1 + u t =
= µ + δ t + Π ( L ) ⋅ Y t −1 + Q − 1 e t
Now everything is done! We have already estimated the coefficients of the VAR model, i.e.
elements of Π matrices, and previously we estimated matrices A and B (which yields Q). The
final step is to plug equations of this system one into the other, then iterating backward in time
and finally obtaining the SMA(∞) representation:
Yt = γ + Φ 0 et + Φ1et −1 + Φ 2 et − 2 + ........
Where Φ are 3x3 matrices, whose elements captures the impulse responses. More precisely, the
analytical expression of the system above is:
π = γ + φ 0 φ 0 φ 0 e + φ 1 φ 1 1
φ 23
t π 21 22 23 πt 21 22 eπt −1 + ........
yt γ y φ31 φ32 φ33 e yt φ31 φ32
0 0 0 1 1
φ331 e yt −1
After specifying matrices A and B, E-views can compute automatically the impulse responses
and yields the following graphs. We concentrate on impulse responses of all variables to a
money shock only.
The impulse response function plots the percentage deviation of a variable from its steady state
against periods of time after the shock. So, for example, a money shock causes money in the first
period after the shock (t=1) to increase almost to 0.017% over its steady state, as we see in the
next graph.
11
ESTONIA
Effects of a monetary policy shock on Money
* Shock 1 represents a money supply shock, while shock 2 represents an interest rate shock.
.04 .2
.1
.02
.0
.00
-.1
-.02
-.2
-.04 -.3
-.06 -.4
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.10
.04
.05
.00 .00
-.05
-.04
-.10
-.08 -.15
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.02
.02
.01
.01
.00
.00 -.01
-.02
-.01
-.03
-.02 -.04
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
12
Effects of a monetary policy shock on Inflation
Response of INFLATION_ES to Structural Response of INFLATION_ES to Structural
One S.D. Shock1 One S.D. Shock1
.06 .2
.04 .1
.02 .0
.00 -.1
-.02 -.2
-.04 -.3
-.06 -.4
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.04
.04
.02
.02
.00
.00
-.02
-.02
-.04
-.04 -.06
-.06 -.08
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.008
.01
.004
.00
.000
-.01
-.004
-.008 -.02
-.012 -.03
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
13
Effects of a monetary policy shock on GDP
.04 .1
.0
.00
-.1
-.04 -.2
-.3
-.08
-.4
-.12 -.5
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.08 .10
.04 .05
.00 .00
-.04 -.05
-.08 -.10
-.12 -.15
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.03 .02
.02 .01
.01 .00
.00 -.01
-.01 -.02
-.02 -.03
-.03 -.04
-.04 -.05
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
14
Effects of a monetary policy shock on Interest Rate
Response of IR_ES to Structural Response of IR_ES to Structural
One S.D. Shock1 One S.D. Shock1
.03 .012
.02 .008
.01
.004
.00
.000
-.01
-.004
-.02
-.008
-.03
-.04 -.012
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.003
.008 .002
.001
.004
.000
-.001
.000
-.002
-.004 -.003
-.004
-.008 -.005
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
15
Comments of the impulse responses for ESTONIA
The impulse response of money and inflation due to a money supply shock is statistically
significant, but is almost zero.
The impulse response of output due to a money supply shock is not statistically
significant, almost zero.
The impulse responses of money, inflation and output, using the interest rate for policy
shocks, are not statistically significant, almost zero.
However, what we could say if the results were significant (in fact we have to include more
observations in our sample to make results robust) :
The results for the other countries, Latvia and Lithuania, are similar. Furthermore, the
impulse responses in all cases are either not statistically significant or almost zero. Hence, the
effects of monetary policy shocks are very weak. (Again the problem is the lack of sufficient
number of observations that would enable us to derive more robust results).
16
Latvia
Effects of a monetary policy shock on Money
.2 .2
.1
.1
.0
.0
-.1
-.1
-.2
-.2 -.3
-.3 -.4
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.04
.04
.03
.02 .02
.01
.00
.00
-.01 -.02
-.02
-.04
-.03
-.04 -.06
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
0.2 .12
0.0 .08
.04
-0.2
.00
-0.4
-.04
-0.6
-.08
-0.8 -.12
-1.0 -.16
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
17
Effects of a monetary policy shock on Inflation
-.01 .00
-.01
-.02
-.02
-.03 -.03
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.08
.010
.06
.005
.04
.000 .02
-.005 .00
-.02
-.010
-.04
-.015 -.06
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.03
.005
.02
.01
.000
.00
-.005 -.01
-.02
-.010 -.03
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
18
Effects of a monetary policy shock on GDP
.3 .20
.15
.2
.10
.1
.05
.0
.00
-.1
-.05
-.2 -.10
-.3 -.15
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.05 .1
.0
.00
-.1
-.05 -.2
-.3
-.10
-.4
-.15 -.5
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.04 .03
.02 .02
.00 .01
-.02 .00
-.04 -.01
-.06 -.02
-.08 -.03
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
19
Effects of a monetary policy shock on Interest Rate
.04
.005
.00
.000
-.04
-.005
-.08
-.010 -.12
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.004 .02
.01
.002
.00
.000
-.01
-.002
-.02
-.004
-.03
-.006 -.04
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
20
Lithuania
Effects of a monetary policy shock on Money
.12 .04
.08 .02
.04 .00
.00 -.02
-.04 -.04
-.08 -.06
-.12 -.08
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.2 .1
.0
.1
-.1
.0
-.2
-.1
-.3
-.2
-.4
-.3 -.5
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.2
.1
.1
.0 .0
-.1
-.1
-.2
-.2 -.3
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
21
Effects of a monetary policy shock on Inflation
.015 .03
.010 .02
.005 .01
.000 .00
-.005 -.01
-.010 -.02
-.015 -.03
-.020 -.04
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.00 .00
-.01
-.01
-.02
-.02
-.03
-.03 -.04
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.02
.08
.01
.04
.00
.00
-.01
-.04
-.02
-.03 -.08
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
22
Effects of a monetary policy shock on GDP
.04
.10
.02
.05
.00
.00 -.02
-.04
-.05
-.06
-.10
-.08
-.15 -.10
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.3 .05
.2 .00
.1 -.05
.0 -.10
-.1 -.15
-.2 -.20
-.3 -.25
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.12 .10
.08
.05
.04
.00
.00
-.05
-.04
-.10
-.08
-.12 -.15
-.16 -.20
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
23
Effects of a monetary policy shock on Interest Rate
.06 .01
.04
.00
.02
.00 -.01
-.02
-.02
-.04
-.06 -.03
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
.02
.02
.01
.01
.00
.00
-.01
-.01
-.02
-.02 -.03
-.03 -.04
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
24