International Transmission of Monetary Policy Shocks The Case of Romanian Economy

1
International Transmission of Monetary Policy
ShocksThe Case of Romanian Economy
The Academy of Economic Studies,
Bucharest DOCTORAL SCHOOL OF FINANCE AND BANKING

• MSc Student ELENA BOJESTEANU
• Supervisor Professor MOISA ALTAR

Bucharest, 2006
2
Contents

• Motivation
• The analysis of comovements using PCA
• Identifying monetary policy shocks using a
  reaction function for the ECB
• The transmission of shocks a SVAR approach
• Concluding remarks and further research

3
Key References

• Two-country models
• - Mundell Fleming (1962)
• - Svensson and van Wijnbergen (1989)
• - Obstfeld and Rogoff (1995)
• - Chari, Kehoe and McGrattan (1997, 2000)
• - Engel and Devereux (2003).
• Empirical Investigations
• - Cushman and Zha (1997)
• - Christiano, Eichenbaum and Evans (1996,
  1998)
• - Betts and Devereux (1999)
• - Bordo and Murshid (2002)
• - Pesaran et al. (2005).

4
Romanian Economy

• The last five years improved performance in
  terms of economic expansion, strengthening
  disinflation, reduction in budget deficit and
  unemployment.
• A small open economy a degree of financial and
  commercial openness exceeding 70 in the last 8
  years. Main trading partner European Union (over
  70 of the total for exports, and more than 60
  of the total imports. Trading currency more than
  60 settlement of exports and imports in euro,
  and approx. 30in US dollar.
• Capital account liberalization schedule. Romania
  is increasingly integrating into world markets
  and more precisely, into European structures.
• Question how do the commercial and financial
  linkages affect the Romanian main economic
  indicators? Is there a comovement in the economic
  variables, a synchronization of business cycles,
  or Romanian is still in an incipient phase of
  integration?

5
Tracing comovements using Principal Component
Analysis

• The method generates a new set of variables,
  called principal components
• PC are orthogonal to each other
• Each principal component (pc) is a linear
  combination of the original variables
• The coefficients of each of the linear
  combinations are called loadings (or weights)
• The first principal component explains the
  greatest amount of the total original variance
• The sum of the pc variances equals the total
  variance of the initial system

6
Tracing comovements using Principal Component
Analysis
GDP growth rates

• The first component loading for Romania is small,
  which shows a lower correlation with the rest of
  the system. Sayek and Selover (2002) state that
  the first principal component might be thought of
  as the business cycle followed by the Western
  nations.
• For the second component, Romania has a much
  higher loading than the rest of the countries.
  The low and positive Romanias loading on the
  first component may be a sign that the
  macroeconomic evolution in this country was in
  general out of step with the rest of the group.

7
Tracing comovements using Principal Component
Analysis
Real interest rates

• A higher degree of comovement in the financial
  sector that in the real one. The first principal
  component explains a substantial portion of the
  behaviour of interest rates across countries and
  can be interpreted as the common element of real
  interest rates.
• An atypical pattern for Romania is not out of the
  question, considering the high correlation with
  the second component.

8
Possible Interpretations for Monetary Policy
Shocks

• Three general strategies for isolating monetary
  policy shocks
• The recursiveness assumption based on the
  estimation of a reaction function for the
  monetary authorities
• - Christiano (1996)
• - Christiano, Eichenbaum and Evans (CEE, 1996,
  1997)
• - Clarida, Gali and Gertler (1997)
• - Cushman and Zha (1997).
• The narative approach
• - Romer and Romer (1989)
• Long-run neutrality of money
• - Pagan and Robertson (1995)

9
Identifying the monetary policy shocks using a
reaction function for the ECB

• An exogenous monetary policy shock, et -
  formalized by CEE (1998) as being the disturbance
  term in an equation of the form
• St f(t) et ,
• where St is the instrument used by the monetary
  authority and f(t) is a linear function that
  captures the policy makers responses to
  variations in different economic variables, as
  they are known at time t.
• An augmented reaction function for the Euro area
• it (1-?)a(1-?)ßEptn (1-?) ?yt ?
  it et .

10
Identifying the monetary policy shocks using a
reaction function for the ECB
Studies estimating the reaction function using
data before EMU Gerdesmeier and Roffia
(2003) Gerlach(2003) Surico (2003) Carstensen
and Colavecchio (2005).

• Data
• ex-post available data
• survey data
• Methodology
• GMM for ex-post available data (a popular
  technique in the rational-expectation context
  (Clarida, 1998)). Problem selection of
  instruments.
• OLS for survey data. Problem constructing the
  data.

11
Identifying the monetary policy shocks using a
reaction function for the ECB

• Ex-post available data
• 199601-200504 from ECB and Eurostat databases
• Interest rates the interbank ON interest rate,
  the 3M EURIBOR, and the 10Y government bond yield
• Price indices annualized HICP and alternatively
  the core inflation (HICP - All items, excluding
  energy, food, alcohol and tobacco)
• Output gap from three measures for potential
  GDP a Hodrick-Prescott filter (the smoothing
  parameter equal to 1600 for quarterly data), a
  linear and a quadratic trend. The three methods
  yield fairly similar results.
• Monetary aggregates M3 a money gap was also
  used (the deviation of money growth from the
  reference value of a constant growth of 4.5 per
  annum)
• Exchange rates nominal and real effective
  exchange rate.

12
Identifying the monetary policy shocks using a
reaction function for the ECB

• Ex-post available data

• GMM. The instrument set includes lagged values
  (up to 4 lags) of the interest rate, inflation
  and output gap. The results are not very
  sensitive in respect to the number of lags used
  as instruments, the J-statistic supports the
  over-identifying restrictions implied by the
  model.
• The standard errors were computed using the delta
  method. The J-statistic reported in the table is
  the minimized value of the objective function,
  p(J), the null hypothesis that the
  overidentifying restrictions are satisfied
  (Hansens J-test).

13
Identifying the monetary policy shocks using a
reaction function for the ECB

• Ex-post available data

14
Identifying the monetary policy shocks using a
reaction function for the ECB

• Survey data
• Sample period 19991-20054
• Quarterly forecasts based only on real-time
  available information
• Solid arguments in favor of using survey data
• - they are more suitable to capture the
  forward-attitude of the policy makers
• - variables (in particular the series for
  output) are only available with lags
• - data are often subject to revisions and it
  may take some quarters before the final series
  are available.
• Inflation based on the data from the Survey of
  Professional Forecasters (SPF), measured by the
  latest available forecast for the current year
• A measure of the state of real economy the
  Economic Sentiment Indicator (ESI). This
  economic index appears to be more closely tied to
  the Governing Councils interest rate decisions
  than other variables capturing real economic
  activity.

15
Identifying the monetary policy shocks using a
reaction function for the ECB

• Survey data
• ESI as a leading indicator for economic activity

Studies estimating the reaction function using
survey data Carstensen and Colavecchio
(2005) Gerdesmeier and Roffia (2005) Gerlach
(2004) Sauer and Sturm (2003).
16
Identifying the monetary policy shocks using a
reaction function for the ECB

• Survey data
• Sample period 19991-20054

• The first two specifications show the results
  without partial adjustment. Although this
  restriction is rejected by the data, the
  estimates correspond to the original Taylor
  coefficients. The constant term is found to be
  statistically insignificant, similar to the
  findings of Carstensen and Colavecchio (2005).
• The equation that best fits the data is
  considered to be 9, with an interest rate
  smoothing and no constant term.

17
Identifying the monetary policy shocks using a
reaction function for the ECB

• Survey data
• Identified monetary shocks

18
Identifying the monetary policy shocks using a
reaction function for the ECB

• Survey data

• The results show a greater weight attached to the
  output gap relative to inflation, a conclusion
  similar to that of the studies using ex-post
  data.
• For some specifications, the constant term is
  found to be statistically insignificant.
• The real time forward-looking specifications of
  the Taylor rule using the SPF forecasts denote a
  stabilizing behavior and provide a better
  description of the actual behavior of the central
  bank.

19
The transmission of shocks a SVAR approach

• Description of the variables. Quarterly data
  comprising
• Inflation rate (pi_ro), calculated using log
  deviation of the CPI from the previous quarter
• Core inflation (core1), CPI all items excluding
  administrated prices
• Real interest rate (rr_ro), the difference
  between BUBOR 3M and the inflation rate
• Real GDP growth rate (d(y_ro))
• Exchange rate appreciation (d(log(er))), as the
  log difference between the quarterly mean of the
  exchange rate and that of the previous period.
• The series are seasonally adjusted using
  TRAMO/SEATS (Demetra). The sample period, due to
  data availability for the European reaction
  function is 19991-20054 (28 observations). CB
  and Eurostat databases.

20
The transmission of shocks a SVAR approach

• Main three methods to identify the pure
  innovations
• The recursive approach (the triangular Choleski
  decomposition)
• The structural approach as advocated by Sims and
  Bernanke
• The long-term restriction approach (the Blanchard
  and Quah decomposition)

21
The transmission of shocks a SVAR approach

• The models
• Model (A1) rr_ro, er, mshock
• Model (A2) core1, er, mshock
• Model (A3) pi_ro, er, mshock
• Model (B) y_ro, rr_ro, er, mshock.
• Isolating pure shocks by Choleski ordering
• Model (A1) mshock ? er ? rr_ro
• Model (A2) mshock ? er ? core1
• Model (A3) mshock ? er ? pi_ro.

22
The transmission of shocks a SVAR approach

• Isolating pure shocks
• The Model (B) relies on an identification scheme
  which assumes that contemporaneously (within a
  quarter), the external monetary shock affects
  only the financial variables (the exchange rate
  and the real interest rate) and not the real
  activity in Romania.

23
The transmission of shocks a SVAR approach

• Tests for selecting the number of lags
• Model (A1) rr_ro, er, mshock
• Model (A2) core1, er, mshock
• Model (A3) pi_ro, er, mshock.

24
The transmission of shocks a SVAR approach

• Tests for selecting the number of lags
• Model (B) y_ro, rr_ro, er, mshock

25
The transmission of shocks a SVAR approach
26
The transmission of shocks a SVAR approach
27
The transmission of shocks a SVAR approach
28
The transmission of shocks a SVAR approach
29
The transmission of shocks a SVAR approach
30
Concluding remarks

• The monetary shocks are isolated by estimating a
  reaction function for the euro area. A more
  appropriate method to identify the policy shocks
  is to use the information set available at the
  moment the decision is made, i.e. survey data.
• The empirical evidence does not support an impact
  of these shocks on the internal variables.
• The number of observations used for the
  estimation may be inappropriate for the analyses
  of monetary transmission, knowing that the
  monetary decisions affect the economy only with
  lags.
• For the major part of the analyzed period,
  Romanias exchange rate regime was managed
  floating, but according to empirical findings and
  to IMF, it was a mixed regime in the form of
  sliding band. The theoretical results for the
  case of floating exchange rate may not hold if
  this assumption is not met.
• Moreover, the capital account has not been fully
  liberalized and the stages with the greatest
  impact on the balance of payments occurred in
  only in 2005. The financial openness is
  questionable before this period.

31
Concluding remarks

• Not European business cycles and monetary
  innovations determine the internal economic
  indicators, but domestic economic and political
  developments.
• During transition period major internal
  disturbances affected the Romanian economy these
  internal shocks include the effects of domestic
  political conflicts, economic and financial
  crises, domestic policy mistakes and so on.
• Apart from the obvious advantages incurred by the
  imminent accession, the unpredictable effect of
  the European monetary policy on the Romanian
  economic variables can trigger integration costs
  not dealt with so far.
• Further research
• Alternative methods for identification of
  monetary policy shocks (using the data-determined
  approach and the Blanchard and Quah
  decomposition).
• The reaction function for the ECB can be obtained
  by employing monthly data, using cubic splines on
  real GDP.
• In order to test the relevance of the theme it is
  useful to analyze the transmission of the
  identified monetary policy shocks in other
  countries except Romania, namely the new EU
  member countries. It can also be tested whether
  there is an asymmetry bween the effects of a
  negative and positive monetary shock.

32
Selected references
33
Selected references
34
Selected references
35
Selected references