Title: The Academy of Economic Studies Doctoral School of Finance and Banking
1The Academy of Economic StudiesDoctoral School
of Finance and Banking
- Monetary Policy Rules Evaluation using a Forward
- Looking Model for Romania
MSc student Murarasu Bogdan Coordinator
Professor Moisa Altar
2I motivate the importance of my topic by the
following remark of
- John Taylor (1998) researchers first build a
structural model of the economy, consisting of
mathematical equations with estimated numerical
parameter values. They then test out different
rules by simulating the model stochastically with
different policy rules placed in the model. One
monetary policy rule is better than another
monetary policy rule if the simulation results
show better economic performance.
3CONTENTS
- Policy evaluation with a forward looking model
- Estimation (calibration) of the model
- Klein algorithm (generalized Schur decomposition)
- Central banks loss function and the optimization
problem - Optimal monetary policy rules
4The forward looking model
- Coats, Laxton, and Rose (2003) argued that in
order to support the policy decisions necessary
to respect a target for inflation, the framework
had to be forward-looking and capable of dealing
with the process of controlling inflation.
(1)
- Another specification of the system includes the
real effective exchange rate in the IS curve.
Taking into consideration that are no great
differences between the two cases regarding the
methodology and even the main results, I will
describe the procedure I follow referring to
first model.
- This model introduces two layers of complexity
1. agents actions depend upon expected future
output and inflation which may cause the
existence of zero or many reduced form equations
2. the system must be solved for simultaneity.
5Estimation vs. Calibration
- Problems
- data are very limited, both in terms of the
coverage and the duration of series - data sample is very short and describes a period
of major structural change in the economy and
major change in policy regimes
These are reasons to expect very imprecise
identification of the parameters from any
estimation.
- Solutions
- I chose a full information method of estimation
(3SLS) in order to solve for simultaneity - after estimation I kept the coefficients that
were statistically or economically significant - I applied a kind of calibration for the
coefficient from the Phillips curve which
is statistically and economically inconsistent
6Data used in estimation
- The model is fitted to quarterly data for the
Romanian economy for 1998Q1 2006Q2, subject to
the restriction that the coefficients of the
policy rule minimize a quadratic loss function.
deviation of inflation from its target
(inflation is measured as a percentage change of
headline CPI, quarter-over-quarter, at annual
rates and is seasonally adjusted using Demetra
(Tramo-Seats))
For the interest rate gap I applied a
Hodrick-Prescott filter to the data and I
computed the gap as a deviation from the trend.
7Structural parameters
- quality of the instruments in 3SLS estimation
8The stability of the coefficients from the two
curves across the interval of variation of
9Structural system is written in Klein format as
(1)
is a vector of predetermined variables
the forward looking or non-predetermined
variables
Reduced form of the system (Klein(2000) algorithm)
10Klein algorithm (generalized Schur decomposition)
- solves systems of linear rational expectations
- the system need to be solved distinctly for the
predetermined variables (or backward-looking in
the language of Klein) and non-predetermined ones
( or forward looking variables) - infinite and finite unstable eigenvalues are
treated in a unified way - preferable from a computational point of view to
other similar numerical methods
11 - for the pair of square matrices from
the equation (1) the orthonormal matrices and the
upper triangular matrices exist
such that
(2) - The generalized eigenvalues of the system are the
ratios where and are
the diagonal elements of and -
- The decomposition matrices can be transformed so
that the generalized eigenvalues are arrayed in
ascending modulus order (stable eigenvalues come
first corresponding to backward looking variables
and unstable come next corresponding to forward
looking variables)
12Solutions
(3)
(4)
13Reduced form
- Now I have the structural system (1) written in
the reduced form as -
- is a
vector of predetermined variables
(5)
- Taking into account equation (5) we can recover
the covariance matrix of structural errors from
the covariance matrix of reduced form errors with
the relationship
14Loss function
- The central bank chooses the values for the
coefficients from the reaction function that
minimize the loss function
- is a matrix of policy weights that represent
the relative importance to the central bank of
stabilizing inflation, output and interest rate
(stabilization objectives). - These weights range between zero and one and sum
to one in order to determine whether the
performance of the policies is sensitive to
policy objectives (represented by the weights
assigned to stabilize inflation, output and
respectively interest rate).
- By minimizing the loss function I also obtain
optimal values for the coefficients of the
reaction function
15Computation of the loss function
Because the reduced form errors are linear
combinations of the serially uncorrelated
structural errors, they are serially uncorrelated.
16Correlograms and serial correlation LM test for
the structural errors
- Tests for no autocorrelation of the residual
(residual from IS curve)
- Tests for no autocorrelation of the residual
(residual from Phillips curve)
17Alternative policy rules
- The interest rate rules proposed by John Taylor
are the most used ones. Taylor Rule with Interest
Rate Smoothing
- Original Taylor Rule (Taylor, 1993) assigns exact
coefficient values that describe Federal Reserve
policy
- Optimal Taylor Rule but chooses the
values for and that minimize the
loss function of the central bank
- Taylor Backward-Looking Rule, where lagged values
of output and inflation replace the current
values of the two variables
- Full State Rule (respond to all, rather than a
subset, of the variables in the state vector)
- Woodford (2002) attributes to Goodhart a simple
rule where the central bank responds only to
deviations of the inflation rate from its target
value and choosing an optimal value
for
- Clarida, Gali and Gertler (1998) suggest that
forecast-based rules are optimal for a central
bank with a quadratic objective function
18(No Transcript)
19Results
- Table 1 reports the policy rule that achieved the
lowest loss level for each set of policy
objective weights considered.
Taylor Rule with Interest Rate Smoothing
Goodhart Rule
Expected Inflation Rule
In the case where NBR gives an important weight
to inflation stabilization, as this is its
primary objective and output represents an
important but secondary objective, the Taylor
Rule with Interest Rate Smoothing is the best
rule to adopt.
20Relative performance of the rules
The figure shows that the Taylor Rule with
Interest Rate Smoothing performs at all times
better than the Taylor Backward Looking Rule.
When the NBR is preoccupied by the stability of
output then it has to respond currently to output
gap and not with a lag.
21Taylor with Interest Rate Smoothing vs. Full
State Rule and Goodhart Rule
The figure shows the superiority of the Taylor
rule against the rule which takes into
consideration the entire state vector. This rule
performs better than the Taylor type rule only
when the stability of inflation is the only
objective of the central bank.
The figure shows that this simple rule can
perform better than the Interest Rate Smoothing
Rule when the output weight is small and also
that the performance of this rule is not
sensitive to weight assigned to interest rate
stabilization.
22Full State Rule vs. Expected Inflation and
Taylor Backward Looking vs. Optimal Taylor
- the central bank should not adopt a policy rule
in which the nominal rate of interest responds
only to changes in the current expectation of
future inflation
- the conclusion is that the central bank performs
better if it conditions its policy on current
rather than lagged economic variables
23Impulse responses to positive demand shock for
four policy rules, namely Taylor Rule with
Interest Rate Smoothing Full State Rule
Backward Looking Rule and Goodhart Rule
- Taylor Rule with Interest Rate Smoothing
- Goodhart(interest conditioned on current
inflation)
24Impulse responses to positive demand shock of
expected inflation and output
Taylor Rule with Interest Rate Smoothing Full State Rule
25It is clear that the Taylor Rule with Interest
Rate Smoothing achieves a much more stable output
gap and inflation, in spite of a relatively small
increase in the nominal interest rate. This is
achieved by credibly committing to a fixed
coefficient rule that conditions the short-term
interest rate to current economic variables and
to lagged interest rate.
Conclusions
- Taylor Rule with Interest Rate Smoothing responds
better to economic conditions in Romania
- A central bank like ours, which takes care mostly
about stabilizing inflation and is concerned
about the economic stability, should control the
interest rate using a Taylor Rule with Interest
Rate Smoothing.
- Paper provides evidence on the practical
importance to a central bank of analyzing the
performance of the commitment mechanism
- In future work, I intend to compare the
performance of fixed coefficients rules to
unconstrained optimal commitment policy and
discretionary policy, two alternatives proposed
by Clarida, Gali and Gertler (1999).
26Reference
- Anderson, E. W., L. P. Hansen, E. R. McGrattan,
and T. J. Sargent, (1996), Mechanics of forming
and estimating dynamic linear economies, in
Amman, H. M., David A., Kendrick, and J. Rust,
eds., Handbook of Computational Economics 1,
Handbooks in Economics 13, Elsevier Science,
North-Holland, Amsterdam, 171-252. - Batini, N., and A. Haldane, (1998),
Forward-Looking rules for monetary policy,
Presented at the NBER Conference on Monetary
Policy Rules. - Batini, N., R. Harrison, and S. P. Millard,
(2002), Monetary policy rules for an open
economy, The Bank of Englands working paper. - Bernanke, B., M. Gertler, and S. Gilchrist,
(1998), The financial accelerator in a
quantitative business cycle framework, NBER
Working Paper 6455. - Blanchard, O. J. and C. M. Kahn, (1980), The
solution of linear difference models under
rational expectations, Econometrica 48, 1305-11. - Chadha, J. S., and L. Corrado, (2006), Sunspots
and Monetary Policy, Centre for Dynamic
Macroeconomic Analysis working papers series. - Clarida, R., J. Gali, and M. Gertler, (1998),
Monetary policy rules and macroeconomic
stability Evidence and some theory, NBER
Working Paper 6442. - Clarida, R., J. Gali, and M. Gertler, (1999a),
The science of monetary policy a new Keynesian
perspective, Journal of Economic Literature
XXXVII, 1661-1707. - (1999b), Inflation dynamics a structural
econometric analysis, Journal of Monetary
Economics 44, 195-222. - Coats, W., D. Laxton, and D. Rose, (2003), The
Czezh National Banks Forecasting and Policy
Analysis System, The Czech National Banks
working paper. - Edwards, S., (2006), The relationship between
exchange rates and inflation targeting
revisited, NBER Working Paper 12163. - Fic, T., M. Kolasa, A. Kot, K. Murawski, M.
Rubaszek, and M. Tarnicka, (2005), ECMOD Model
of the Polish Economy, The National Bank of
Polands working paper. - Giannoni, M. P., (2006), Robust Optimal Policy
in a Forward-Looking Model with Parameter and
Shock Uncertainity, NBER Working Paper 11942.
27- Givens, G., 2002, Optimal monetary policy
design solutions and comparisons of commitment
and discretion, University of North Carolina. - Hansen, L. P. and T. J. Sargent, (1980),
Formulating and estimating dynamic linear
rational expectations models, Journal of
Economic Dynamics and Control 2, 7-46. - Klein, P., (2000), Using the generalized Schur
form to solve a multivariate linear rational
expectations model, Journal of Economic Dynamics
and Control, 24, 1405-23. - Kydland, F. E., and E. C. Prescott, (1977),
Rules rather than discretion The inconsistency
of optimal plans, Journal of Political Economy,
85, 473-491. - Lubik, T. A., and F. Schorfheide, (2003),
Computing sunspot equilibria in linear rational
expectations modelsJournal of Economic Dynamics
Control 28, 273 285. - Ljungqvist, L. and T. J. Sargent, (2000),
Recursive macroeconomic theory, MIT Press,
Cambridge, MA. - Mohanty, M. S., and M. Klau, (2004), Monetary
policy rules in emerging market economies issues
and evidence, BIS working paper no. 149. - Onatski, A., and N. Williams, (2004), Empirical
and Policy Performance of a Forward-Looking
Monetary Model, Princeton University working
paper. - Przystupa, J., and E. Wrobel, Looking for an
Optimal Monetary Policy Rule The Case of Poland
under IT Framework, National Bank of Poland. - Salemi, M. K., (1995), Revealed preference of
the Federal Reserve using inverse-control theory
to interpret the policy equation of a vector
autoregression, Journal of Business and Economic
Statistics 13, 419-433. - Soderlind, Paul, (1999), Solution and estimation
of RE macromodels with optimal policy, European
Economic Review, 43, 813-23. - Soderlind, Paul, (2003), Lectures Notes for
Monetary Policy (PhD course at UNSIG),
University of St. Gallen and CEPR. - Svensson, L. E. O., (2000), Open-economy
inflation targeting, Journal of International
Economics, 50, 155-83. - Taylor, John B., (1993), Discretion versus
policy rules in practice, Carnegie-Rochester
Conferences Series on Public Policy, 39, 195-214. - Taylor, John B., (1998), Applying academic
research on monetary policy rules an exercise in
translational economics, The H. G. Johnston
Lecture, Macro, Money, and Finance Research Group
Conference, Durham University, Durham England,
revised. - Woodford, M., (2002), Interest and Prices,
Princeton University Press.