Bucharest University of Economics Doctoral School of Finance and Banking DOFIN

1
Bucharest University of EconomicsDoctoral School
of Finance and BankingDOFIN

• Policy Mechanism
• Transmission Channels in Romania

Supervisor Professor Dr. Moisa Altar MSc
Student Ion Savulescu
Bucharest July 2008
2
Contents

• The objectives of the dissertation paper
• The actual stage of research in the field of
  substantiating the monetary policy using VAR
  econometric models
• The theoretical substantiation of the
  transmission channels for the monetary policy and
  the justification of the methodology and
  techniques.
• Utilised Data and processing methodology
• The results I obtained
• Conclusions

3
1. The objectives of the dissertation paper

• The identification of the monetary transmission
  mechanism main features in Romania, using
  econometric models (VAR methodology)
• Using the estimated VAR models (including
  structural VAR) I pursued the identification of
  the monetary policy transmission channels and
  also of a way of modeling the money demand

4
2. The actual stage of research in the field of
substantiating the monetary policy using VAR
econometric models

• During the evolution of the economic science the
  formulation of the first transmission mechanism
  for the monetary policy belongs to J. M. Keynes
  Specification of a structural model of the
  effect of monetary policy over the economic
  activity
• On the other side, the monetarists backed the
  second approach, by specifying a model with
  reduced form and the analysis of the relation
  between the levels of the money supply and that
  of the economic activity, the estimation
  correlation coefficient between the two variables
  (Milton Friedman the promoter)

5
3. The theoretical substantiation of the
transmission channels for the monetary policy and
the justification of the methodology and
techniques

• Monetary policy leads to strong, rapid and
  generalized effects over some variables like
  prices and production, these actually being the
  main objectives of this type of actions

Change in the monetary policy instrument
Interest rates Exchange rates
Alterations of the economic agents and
households behavior
Alterations of the financial assets prices
Deviations from the equilibrium values of
production and unemployment
Wages and prices adjustment to a new equilibrium
6

• Many economists agree with the claim according to
  which the effects of monetary policy over the
  production begin to appear after some time and
  are effects on a relatively short term,
  production receding on long term to its natural
  level.
• The main monetary transmission channels are
• The interest rate channel
• The exchange rate channel
• The assets prices channel
• The credit channel
• The expectations channel

7
4. Utilised Data and processing methodology
Symbol Description (monthly)
BZ Rezerv Money, in mil. Lei
BZR Real BZ, CPI deflated, base 101990
CRNG Total credit to Non-Governments, in mil. Lei
CRNGR Real CRNG, CPI deflated, base 101990
CS Exchange rate (lei/euro)
CSR Real exchange rate, CPI deflated, base 011990
DA Landing rate for non-bank customers
DAR Real landing rate for non-bank customers
DPMML Monetary Policy Interest Rate
8
DPMMLR Real Monetary Policy Interest Rate
DP Deposit rate to non-bank customers
DPR Real DP, CPI deflated, base 101990
IPCL Inflation CPI
IPCX Inflation, CPI, chain, base 101990
IPCXLOG Inflation, CPI, chain, in logs
IPPIX PPI, base 101990
IPPIL PPI chain, base 101990
IPPXLOG PPI in logs
M1 M1
M1R Real M1, CPI deflated, base 101990
M2 M2
M2R Real M2, CPI deflated, base 101990
SC NBRs Reference Rate
9
SCR Real NBRs Reference Rate
SMNB Nominal gross average monthly wage
SMRB Real gross average monthly wage
VPIX Industrial output variation rate, base 021990
VPIL Industrial output variation chain
10
(No Transcript)
11
The used methodology

• I studied the seasonality using U.S. Census
  Bureau X-12 monthly seasonal adjustment method
• I also studied the stationarity of the series
• Granger-causality test
• Regression equation of the industrial
  productions variation (VIP) on the main
  variables in the analyzed group
• Limited model of unrestricted VAR, with three
  endogenous variables (VPIX, CRNGR and M1R) and
  four exogenous (BZR, DP, IPCX and SMRB), with a
  number of 6 lags.
• 14 unrestricted VAR models with 7 variables and
  six lags
• 1 SVAR model
• Cointegration test

12
5. The results I obtained

• 5.1 Granger causality tests
• I applied Granger causality tests in two steps.
  In the first stage I applied the test on the
  entire set presented in section 4. Using the
  results from this step, I selected a group of 12
  variables on which I applied the Granger
  causality test again.
• By processing the results from the second step
  (the elimination of the pairs with the
  probability of the hypothesis over the threshold
  of 5, the grouping of the remaining variables in
  cause variables), I was able to make the
  following observations regarding the causality
  relations between the studied variables

13
(No Transcript)
14
(No Transcript)
15

• There are causality relations between the
  majority of the variables in the study (BZR, M1R,
  M2R) and the non-governmental credit, which seems
  to indicate the presence of the credit channel in
  the monetary policy mechanism
• The exchange rate has an influence well showed by
  the tests results both on the monetary variables
  (BZR, M1R, SC) and on the inflation (IPCX) and
  over the interest rates in use at the commercial
  banks (DAR, DPR) this seems to indicate the
  channel of the exchange rate is working
• The inflation (IPCX) influences all the monetary
  variables (except for the NBRs Reference rate)
  and also the variables of the economys real
  sector (VPIX, SMBR, IPPX), the commercial banks
  interest rates (DAR and DPR) and exchange rate
  (CSR). I believe that this observation can be
  considered a modest argument for the appositeness
  of choosing the inflation targeting as an
  objective of the monetary policy.

16
5.2 Regression equation of the industrial
productions variation (VIP) on the main
variables in the analyzed group
17
I resumed the regression, eliminating the
variables that had an insignificant influence
18
From these regressions I was able to make the
following observations

• -There is an important influence of the
  industrial productions previous value, of the
  real governmental credit and of the monetary
  supply in a restricted sense and a smaller
  influence of the real monetary base
• I
• - In the case of the credit and in that of the
  monetary supply, the contemporaneous has an
  inverted direction in comparison to that of the
  previous period.

19
5.3 Unrestricted VAR model

• On the ground of previous results I computed an
  unrestricted VAR on three endogenous variables
  (VPIX Industrial output variation rate, CRNGR
  Total real credit to Governments and M1R Real
  M1) and four exogenous variables (BZR Real
  Reserve Money, DP Real deposit rate to non-bank
  customers, IPCX Inflation, CPI, chain and SMRB
  Real gross average monthly wage) and with a
  number of six lags.
• I have, thereby, obtained the graphs representing
  the endogenous variables responses to to the
  shocks of a standard error of each of these.

20
(No Transcript)
21
From these graphs we ca observe

• The positive reaction of the industrial
  productions variation in response to an impulse
  on the non-governmental credit as well as the
  fact that the productions stabilization is being
  done at a higher level
• An impulse on the monetary supply leads, in the
  first part, to a negative reaction of the
  industrial production followed by waving movement
  where the positive components are dominated and
  the amplitude is declining. The shock is desorbed
  after 6 7 periods (months) the industrial
  production reversing to the previous level.

22
14 unrestricted VAR models

• Upon the estimation and analysis of a long series
  of VAR models I kept 14 of those whose structure
  is presented below. From among those I selected
  three models that I presented in the thesis both
  as structure and as the result of the usage of
  the functions impulse-response and of the
  decomposition of that option/variation.

23
(No Transcript)
24
Following, I will present one of the models I
used

• The variables

25
The graphs for all variables responses in the
model to the impulses coming from each of these
are
26
From the Analysis of these graphs it can be
inferred that

• The positive variation of the non-governmental
  credit to the shocks on the monetary policy
  variables (monetary base and monetary supply in a
  restricted sense). The stabilization of the
  credit following a shock on the monetary supply
  is being achieved at a higher level of the
  credit
• An ample response of all the models variables to
  the shock on the consumer price index
  (inflation)
• The shock on inflation has a negative effect on
  the variation of the industrial production and
  its stabilization is being achieved at a lower
  level
• The consumer price index is quite sensible to the
  shocks on the majority of the analyzed variables
• A persistent waving movement (more than 20
  periods), with dominant positive components, is
  caused by the exchange rate on the inflation
  index (IPC).

27
The responses of the consumer price index to one
standard deviation shocks on the variables in
model 1 are portrayed in the following graph.
28
The varince decomposition of the consumer price
index is
29
The response of the models variables to a
standard deviation shock on the consumer price
index
30
5.4 Structural VAR (SVAR)

• The main purpose in the estimation of the SVAR
  models is to obtain an un-recursive
  orthogonalization of the error terms for the
  impulse-response analysis. This alternative to
  the recursive Colesky orthogonalization requires
  the user to impose sufficient restriction in
  order to identify the orthogonal components of
  the error terms
• In this paper I made an SVAR model, only with
  short term restrictions, using the following VAR
  model

31
(No Transcript)
32

• I identified and introduced 70 restrictions by
  fixing 70 elements of the matrixes that needed to
  be estimated (the structural form matrixes of the
  autoregressive vector). Using the procedure
  Estimate Structural Factorization in EViews, I
  estimated the SVAR model.
• Analyzing the impulse-response function from the
  estimated model, one can notice an ample effect
  on the systems variables determined by the shock
  on the exchange rate.

33
(No Transcript)
34
5.5 Cointegration tests

• The purpose of these tests is to determine
  whether a group of non-stationary variables are
  cointegrated. If for a group of time series, of
  which one or more are not stationary, a
  stationary linear combination is identified, one
  can say the series of the group are cointegrated.
  The stationary linear combination is called
  cointegration equation and can be viewed as a
  long-term equilibrium relation between the
  variables. The presence of the cointegration
  relation is the basis for the Vector Error
  Correction (VEC) models.
• I applied the cointegration test for the
  unrestricted VAR model presented in section 5.4.

35

• The results of the test show the following
• According to the trace test
• For a 5 significance level there are 4
  cointegration equations
• For a 1 significance level there are 3
  cointegration equations
• According to the max eigenvalue test, there are
  3 cointegration equations at both the 1 and the
  5 levels

36
A synthesis for the results of the cointegration
test is showed below.
37
(No Transcript)
38
6. Conclusions

• Bank credits affect the actual activity in the
  economy (represented in the study herein by the
  industrial production and the average gross
  salary). On its part, the credit is affected on
  a short term by the monetary policy variables.
  I consider these elements to be a proof of the
  existence and functioning of the bank credit
  channel as one of the main mechanism for the
  monetary policy diffusion in Romania.
• Consumer price index (the inflation) is a
  variable very sensitive to the shocks and
  influences of the monetary variables, but also,
  of the macroeconomic variables. I consider this
  modest emphasize on the inflation manifestation
  on the current Romanian economy, accomplished by
  the study carried out in this paper, to be a
  justification for the appropriateness of aiming
  to choose target inflation as goal of the
  monetary policy in Romania.

39

• The exchange rate is another channel through
  which the monetary policy has been diffused in
  the Romanian economy during the analyzed period.
  Exchange rate variation is also highly
  influenced by the domestic innovations and the
  monetary shocks. Considering the domestic
  innovations as main indicator of the forecasts,
  we notice that these represent the main
  determinant, on a short term, of the exchange
  rate evolution.
• The test performed on the patterns developed and
  presented in the paper confirm the assessment of
  many Economists, according to whom, the monetary
  policy effect on production occurs after a long
  period of time and are effects on a relatively
  short term, the production retrieving its natural
  level on a long term

40

• Thank you for your attention!