2nd Workshop on Macroeconomic Policy Research, Budapest, October 2-3, 2003
Timo Wollmershäuser
Institute for Economic Research
at the Universityof Munich
Should Central Banks React to Exchange Rate Movements?
Slide 2
Empirical observation
Central banks respond with their interest rate instrument to exchange rate movements:
• Evidence from VARs:- Clarida and Gertler (1997): Bundesbank- Brischetto and Voss (1999), Dungey and Pagan (2000): Australia
• Evidence from direct estimation of monetary policy rules:- Clarida, Gali and Gerler (1998): Bundesbank, Bank of Japan, Bank
of England- Gerlach and Smets (2000): Reserve Bank of New Zealand, Bank of
Canada- Ades, Buscaglia and Masih (2002): Chile, Israel, South Africa, the
Czech Republic and Mexico
Slide 3
Results from normative policy-evaluation studies
Taylor (2001): “Research to date indicates that monetary policy rules that react directly to the exchange rate, as well as to inflation and output, do not work much better in stabilizing inflation and real output and sometimes work worse than policy rules that do not react directly to the exchange rate”:
• Small improvement of the macroeconomic performance of a central bank’s interest rate policy: Ball (1999), Svensson (2000), Batini et al. (2001), Leitemo and Söderström (2001);
• Deterioration: Côté et al. (2002);
• Mixed results: Taylor (1999).
Slide 4
Outline of the presentation
I. Reproducing the results from simulation studies
II. Explaining the results from simulation studies
III. Modifying the simulation studies in order to provide a rationale for the empirical results
Slide 5
I. Reproducing the results from simulation studies
Normative policy-evaluation approach
Baseline macro-
econometric model
of the open economy
goal variables
exogenousdisturbances
monetary policy rule
performance of the policy rules
Slide 6
The baseline macroeconometric model
• Phillips curve:
• IS curve:
• Uncovered interest parity:
• Real exchange rate:
• stochastic disturbances- foreign variables:- risk premium (UIP) shock:- variances of the white-noise shocks:
t 1 t t t t 1 t 10.4y 0.2 q q
yt 1 t t t t t 1y 0.8y 0.6 i 0.2q
f st t t t 1 t ti i E s s u
ft t 1 t t 1 t tq q s s
f f f ft t 1 t ti 0.3i , 0
s s st t 1 tu 0.3u
y f st t t tVar Var Var Var 1
Slide 7
A battery of simple policy rules
open economy rulesTaylor rule
R 2 t t y t q ti f f y f q
R 3 t t y t q ( 1 ) t 1i f f y f q
R 4 t t y t q ti f f y f q
R 5 t t y t s ti f f y f s
R 6 t t y t q t q ( 1 ) t 1i f f y f q f q
R 7 t t y t s ti f f y f s
R1 t t y ti f f y
two categories of rules
Slide 8
Optimised simple rules in the baseline model
optimisation: yy q q ( 1)
2t T 2t 0 t t t
f ,f ,f ,f t 0
min Loss E y
s t r u c t u r e o f t h e r u l e a b s o l u t e
l o s s V a r ( t ) V a r ( y t )
R 1 : t t ti 1 . 9 0 1 . 2 5 y 5 . 0 5 2 . 6 8 2 . 3 8
R 2 : t t t ti 2 . 2 3 1 . 5 6 y 0 . 2 7 q 5 . 0 0 2 . 6 8 2 . 3 2
R 3 : t t t t 1i 1 . 8 8 1 . 3 5 y 0 . 1 6 q 5 . 0 1 2 . 6 3 2 . 3 8
R 4 : t t t ti 2 . 1 7 1 . 6 9 y 0 . 2 6 q 4 . 9 4 2 . 6 0 2 . 3 3
R 5 : t t t ti 1 . 9 1 1 . 6 9 y 0 . 2 6 s 4 . 9 4 2 . 6 0 2 . 3 3
R 6 : t t t t t 1i 2 . 2 9 1 . 7 8 y 0 . 3 6 q 0 . 2 3 q 4 . 9 3 2 . 6 2 2 . 3 1
R 7 : t t t ti 1 . 9 0 1 . 2 5 y 0 s 5 . 0 5 2 . 6 8 2 . 3 8
Slide 9
II. Explaining the results from simulation studies
Underlying exchange rate model: uncovered interest parity
The determinants of the spot exchange rate are
• the foreign nominal interest rate;
• the domestic nominal interest rate (the policy instrument);
• the UIP disturbance (risk premium shocks).
f st t t j t j t j
j 0
s E i i u
Slide 10
Optimised policy rules under a perfectly holding UIP condition and constant foreign interest rates
1. No informational gain from responding to contemporaneous exchange rate movements: the interest rate is the only determinant of the exchange rate.
2. However, small gain from commitment to an inertial policy rule: the reaction to lagged exchange rate movements can be regarded as substitute for interest rate smoothing.
s t r u c t u r e o f t h e r u l e a b s o l u t e l o s s V a r ( t ) V a r ( y t )
R 1 : t t ti 1 . 9 1 1 . 2 7 y 4 . 9 3 2 . 6 2 2 . 3 1
R 2 : t t t ti 1 . 9 1 1 . 2 7 y 0 q 4 . 9 3 2 . 6 2 2 . 3 1
R 3 : t t t t 1i 1 . 8 5 1 . 3 5 y 0 . 2 0 q 4 . 8 8 2 . 5 7 2 . 3 1
R 4 : t t t ti 2 . 1 2 1 . 6 1 y 0 . 2 2 q 4 . 8 8 2 . 5 7 2 . 3 1
R 5 : t t t ti 1 . 9 0 1 . 6 1 y 0 . 2 2 s 4 . 8 8 2 . 5 7 2 . 3 1
R 6 : t t t t t 1i 2 . 4 0 1 . 8 9 y 0 . 4 5 q 0 . 2 4 q 4 . 8 8 2 . 5 7 2 . 3 1
t t t t 1i 1 . 7 9 1 . 3 9 y 0 . 1 8 i 4 . 8 8 2 . 5 7 2 . 3 1
Slide 11
III. Modifying the simulation studies in order to provide a rationale for empirical results
1. Introducing the idea of exchange rate uncertainty
2. Discussing the consequences of exchange rate uncertainty
Slide 12
Exchange rate uncertainty
• The central bank considers one specific exchange rate model to be most likely (the supposedly “true” exchange rate model):baseline model with uncovered interest parity with known
stochastic properties ( and )which is used – as before – for deriving the policy rules.
• There are, however, a range of alternative specifications according to which the exchange rate may behave.These specifications occur with an unknown probability
distribution.They incorporate various elements that are typically
employed in the literature to explain deviations from uncovered interest parity.
s s st t 1 tu 0.3u s
tVar 1
Slide 13
Modelling exchange rate uncertainty
s s st s t 1 tu u
qt i t t tq i
f qt r t 1 t 1 q t 1 tq r r q
f st t t 1 t 1 t t ts E s 1 s i i u
qt 1 q t t 1q q
Slide 14
The set-up of the uncertainty evaluation
IS curve+
Phillips curve+
U1/U2/U3/U4/U5/U6
goal variables
monetary policy rules R1to R6 derived
from the baseline model
exogenousdisturbances
(demand & supply shocks as in the baseline model, exchange rate
shocks accordingto U1 to U6)
performance of the policy rules
Slide 15
Performance of monetary policy under U1
0 0.2 0.4 0.6 0.8 1
4
6
8
10Exchange rate uncertainty 1
s
f s s s st t t t 1 t t t s t 1 ti i E s s u with u u
Los
s
Slide 16
Performance of monetary policy under U2
0 1 2 3 4
4
6
8
10Exchange rate uncertainty 2
i
qt i t t tq i
Los
s
Slide 17
Performance of monetary policy under U3
0 1 2 3 4
4
6
8
10Exchange rate uncertainty 3
r
f qt r t 1 t 1 t 1 tq r r 0.5q
Los
s
Slide 18
Performance of monetary policy under U4
0 0.2 0.4 0.6 0.8 1
4
6
8
10Exchange rate uncertainty 4
f st t t 1 t 1 t t ts E s 1 s i i u
Los
s
Slide 19
Performance of monetary policy under U5
0 0.2 0.4 0.6 0.8 1
4
6
8
10Exchange rate uncertainty 5
Los
s
Slide 20
Performance of monetary policy under U6
0 0.2 0.4 0.6 0.8 1
4
6
8
10Exchange rate uncertainty 6
q
qt 1 q t t 1q q
Los
s
Slide 21
Summary of the results
Exchange Rate Uncertainty
Best Performing
Policy RuleSecond Best
Performing Policy Rule
U1 R2 R6
U2 R6 R2
U3 R4 and R5 R6
U4 R2 and R6 R4 and R5
U5 R6 R4 and R5
U6 R2 R6
Slide 22
The quest for robustness in monetary policy
1. A policy rule with a feedback from the lagged and the current real exchange rate is superior to a Taylor-type rule according to which the central bank only responds to movements in domestic goal variables.
2. Such a policy rule that performs best across a range of structural models is called a robust policy rule since it best possibly insulates the economy from the negative consequences of uncertainty about the true exchange rate model.
3. Reacting to exchange rate movements reflects the monetary policymaker‘s quest for robustness in world which is surrounded by a high degree of uncertainty about the true determination of the exchange rate.
Slide 23
Conclusion
4. The reason why normative policy-evaluation studies typically come to the result that responding to exchange rate movements is redundant stems from their assumption of a well-defined and reliable relationship between the interest rate and the exchange rate.
5. Relaxing this assumption and allowing for uncertainty about this relationship provides an economic rationale for the empirically observable interest rate response to exchange rate movements.
Slide 24
T H A N K
Y O U