FX intervention and monetary policy design: a market … · Show importance of both portfolio-balance/ signaling channels. I FX interventions in a NK-DSGE setup: Benes et al. (2013),

Gerencia deEstudios Economicos

FX intervention and monetary policy design: amarket microestructure analysisCarlos Montoro (BIS) & Marco Ortiz (BCRP & LSE)

Presented by:Marco Ortiz

Fourth BIS CCA Research ConferenceSantiago de Chile, Chile

[email protected]

All views expressed in this presentation arethose of the author and do not necessarilyrepresent the views of the Central ReserveBank of Peru.

Marco Ortiz April 2013 1/22


Table of Contents

Introduction

The Model

Results

Heterogeneous Information

Conclusions



MOTIVATION

I Many central banks (EMEs/AEs) have reacted with FX (sterilised)interventions to capital inflows.

FX intervention : 2009 - 2012

(as a % of average FX reserve minus gold)



MOTIVATION

Questions that need to be addressed

I How sterilised intervention affects the transmission mechanism ofmonetary policy?

I Which channels are at work (portfolio/signaling channel)?

I Are there benefits for intervention rules?

I What should be the optimal monetary policy design?



What other authors have done? (1)

I Messe & Rogoff (1983): random walk predicts exchange rates betterthan macroeconomic models.

I Lyons (2001): ”the exchange rate determination puzzle”.I FX microstructure. Evans & Lyons (2002) and others: short-run

exchange rate volatility is related to order flow.I Information heterogeneity. Bacchetta & van Wincoop (2006):

exchange rates in the short run closely related to order flow (little withfundamental).

I Vitale (2011): extends Bacchetta & van Wincoop (2006) to introduceFX intervention. Show importance of both portfolio-balance/ signalingchannels.

I FX interventions in a NK-DSGE setup: Benes et al. (2013), Vargas etal. (2013), Escude (2012).



What do we do?

1) We extend an SOE New Keynesian model, including:

I A market of risk averse FX dealers.

I An explicit role for exchange rate volatility.

I the interaction of FX intervention with monetary policy.

I Extension: information heterogeneity across FX dealers.

2) We extend Townsend (1983) / Bacchetta & van Wincoop (2006)method to solve a DSGE model with heterogeneous information.



What do we find?

FX intervention...

I strong interactions between FX intervention and monetary policy,

I the source of exchange rate movements matters for the effectivenessof interventions,

I rules can make FX interventions more effective as a stabilisationinstrument (expectations channel),

I overall, the control over the exchange rate variance reduces theimportance of non-fundamental shocks in the economy,

I this results are still valid under heterogeneous information, whereinterventions can restore the connection with observed fundamentals.



The model (1)

I Standard NK-SOE DSGE model with an FX market run by risk aversedealers.

I Each dealer d receive FX market orders from households, foreigninvestors and the central bank.

I Dealers are short-sighted and maximise:

max−Edt e−γΩdt+1

where Ωdt+1 = (1 + it)B

dt + (1 + i∗t )St+1B

d∗t is total investment after

returns.



The model (2)

I The demand for foreign bonds by dealer d:

Bd∗t =

i∗t − it + Edt st+1 − stγσ2

where σ2 = vart (∆st+1) is the time-invariant variance of thedepreciation rate.



The model (3)

I Aggregating over dealers: modified UIP (similar to B&vW 2006)

Etst+1 − st = it − i∗t + γσ2($∗t +$∗,cbt )

Et : average rational expectation across all dealers.$∗t : capital inflows$∗,cbt : CB intervention (FX sales).

I In our baseline case, under perfect information, Et(x) = Et(x).



Monetary authority (1)

I Central bank implements monetary policy by setting the nominalinterest rate according a Taylor rule:

ıt = ϕπ(πt) + εintt

I Three different strategies of FX interventionI Pure discretional intervention:

$∗cbt = εcb1t

I Exchange rate rule:

$∗cbt = φ∆s∆st + εcb2t

I Real exchange rate misalignments rule:

$∗cbt = φrerrert + εcb3t



Other equations of interest

I Aggregate demand

yt = φC(ct) + φX(xt)− φM (mt)

I Aggregate supply

πt = ψπHt + (1− ψ)πMt

πHt = κHmct + βEtπHt+1

I Current account

φ$(bt − β−1bt−1

)= tdeft + yt − φCct + φ$/β (it−1 − πt)



Perfect Information: Results (1) - Rules vs. Discretion

1 2 3 4 50

1

2

3

4

5

6x 10

−3 FX Intervention

Dep. ruleDisc.

1 2 3 4 5−10

−5

0

5x 10

−3 Ex. Rate Depreciation Rate

(a) Int. Rule 1

1 2 3 4 50

1

2

3

4

5

6

7x 10

−3 FX Intervention

RER ruleDisc.

1 2 3 4 5−7

−6

−5

−4

−3

−2

−1

0

1x 10

−3 Real Exchange Rate

(b) Int. Rule 2Marco Ortiz April 2013 13/22


Results (2) - Interaction with Monetary Policy

Figure: Reaction to a 1% Monetary Policy Shock - Rules vs. No Intervention

1 2 3 4 5−4

−3

−2

−1

0

1x 10

−3 GDP

No Int. Dep. ruleRER rule

1 2 3 4 5−20

−15

−10

−5

0

5x 10

−4 Inflation

1 2 3 4 5−5

0

5

10x 10

−3 Interest rate

1 2 3 4 5−0.01

−0.005

0

0.005

0.01Depreciation rate

1 2 3 4 5−10

−5

0

5x 10

−3 Real exchange rate

1 2 3 4 5−5

0

5x 10

−4 Exports



Results (3) - Contribution of Shocks under FX Intervention

Importance of fundamental/non−fundamental on Ex. Dep. Var.

Fx Intervention parameter on Dep. Rule

% (

varia

nce

deco

mpo

sitio

n)

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1Non Fundamental Cap. FlowsForeign Interest RateOther Shocks

(a) ∆s rule

Importance of fundamental/non−fundamental on Ex. Dep. Var.

Fx Intervention parameter on RER Rule%

(va

rianc

e de

com

posi

tion)

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.20

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1Non Fundamental Cap. FlowsForeign Interest RateOther Shocks

(b) RER rule

Figure: Variance Decomposition



Results (4) - Effect of FX Intervention Rules

0

0.2

0.4

0.6

0.8

1

1.2

1.4

No Int. φ∆ s = 0.25 φ∆ s

= 0.5 φRER

= 0.15 φRER

= 0.30

Ratio of volatilities: non−fundamental capital flows (no intervention = 1)

FX intervention rule

Rat

io o

f vol

atili

ties

InflationProduction

(a) ω∗ shock

0

0.2

0.4

0.6

0.8

1

1.2

1.4

No Int. φ∆ s = 0.25 φ∆ s

= 0.5 φRER

= 0.15 φRER

= 0.30

Ratio of volatilities: foreign interest rate shock (no intervention = 1)

FX intervention ruleR

atio

of v

olat

ilitie

s

InflationProduction

(b) i∗ shock



Results (5) - Effect of FX Intervention Rules (2)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

No Int. φ∆ s = 0.25 φ∆ s

= 0.5 φRER

= 0.15 φRER

= 0.30

Ratio of volatilities: foreign output shock (no intervention = 1)

FX intervention rule

Rat

io o

f vol

atili

ties

InflationProduction

(c) y∗ shock

0

0.2

0.4

0.6

0.8

1

1.2

1.4

No Int. φ∆ s = 0.25 φ∆ s

= 0.5 φRER

= 0.15 φRER

= 0.30

Ratio of volatilities: foreign inflation shock (no intervention = 1)

FX intervention ruleR

atio

of v

olat

ilitie

s

InflationProduction

(d) π∗ shock



Heterogeneous information structure (1)

I Foreign investor exposure equals average + idiosyncratic term:

$d∗t = $∗t + εdt

I $∗t is unobservable and follows an AR(1) process

$∗t = ρ$$∗t−1 + ε$

∗t

where ε$∗

t ∼ N(0, σ2

$∗). The assumed autoregressive process is

known by all agents.



Heterogeneous information structure (2)

I Now dealers observe past and current fundamental shocks, whilealso receive private signals about some future shocks.

I At time t dealer d receive a signal about the foreign interest rate oneperiod ahead:

vdt = i∗t+1 + εvdt

where εvdt ∼ N(0, σ2

vd

)is independent from i∗t+1 and other agent’s

signals. We also assume that the average signal received byinvestors is i∗t+1, that is

∫ 10 v

dt dd = i∗t+1.

I For the solution we extend Townsend (1983) and Bacchetta and vanWincoop (2006) to a DSGE model. here .



Results (6) - The Effects of HI

0.02 0.04 0.06 0.08 0.100.9

0.95

1

1.05

1.1

1.15

1.2

1.25

1.3

Std. Dev. of Unobservable Shock

Res

pons

e to

i*(t

+1)

sho

ck

(e) Reaction to a i∗t+1 - CK

0.002 0.004 0.006 0.008 0.0100.01

0.012

0.014

0.016

0.018

0.02

0.022

0.024

Std. Dev. of Unobervable Shocks

Rea

ctio

n to

i*(t

+1)

Sho

ck

No InterventionDep. Rule (varphi = 0.25)Dep. Rule (varphi =0.50)

(f) Reaction to a i∗t+1 - HI

0.02 0.04 0.06 0.08 0.10−1.3

−1.25

−1.2

−1.15

−1.1

−1.05

−1

−0.95

−0.9

−0.85


Res

pons

e to

i*(t

+1)

sho

ck

(g) Difference (HI-CK)

0.02 0.04 0.06 0.08 0.10−0.48

−0.46

−0.44

−0.42

−0.4

−0.38

−0.36

−0.34

−0.32

−0.3


Res

pons

e to

uno

bser

vabl

e sh

ock

(h) Reaction to a ω∗t - CK

0.002 0.004 0.006 0.008 0.010−0.9

−0.8

−0.7

−0.6

−0.5

−0.4

−0.3

−0.2


Rea

ctio

n to

uno

bser

vabl

e sh

ock


(i) Reaction to a ω∗t - HI

0.002 0.004 0.006 0.008 0.010−0.4

−0.35

−0.3

−0.25

−0.2

−0.15

−0.1

−0.05

0

0.05


Rea

ctio

n to

i*(t

+1)

Sho

ck


(j) Magnification (HI-CK)Marco Ortiz April 2013 20/22


Results (7) - FX Intervention under HI

0 0.25 0.500

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Intervention Rule Coefficient

R2

R2 Unobservable R2 Fundamental

(k) ∆s rule

0 0.25 0.500

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Intervention Rule Coefficient

R2

R2 UnobservableR2 Fundamental

(l) RER rule

Figure: Regression of ∆st on unobservable and fundamental shocks



Conclusions

I We present an alternative model of exchange rate determination ingeneral equilibrium that can be useful:

I to explain puzzles in the new international economy literature.I for policy analysis (central banks).

I Our results of FX intervention in general equilibrium:I Effective as an instrument in face of financial shocks, but not so much

in face of real shocks or nominal external shocks;I FX intervention rules can have stronger stabilisation power than

discretion as they exploit the expectations channel;I with heterogeneous information, FX intervention can help restore

connection between exchange rate and fundamentals.

I Additional exercises: welfare analysis (eg welfare frontiers fordifferent rules), robustness exercises, informative content ininterventions, interventions under noisy/imperfect information.



FX intervention and monetary policy design: amarket microestructure analysisCarlos Montoro (BIS) & Marco Ortiz (BCRP & LSE)

Presented by:Marco Ortiz

Fourth BIS CCA Research ConferenceSantiago de Chile, Chile

[email protected]

All views expressed in this presentation arethose of the author and do not necessarilyrepresent the views of the Central ReserveBank of Peru.



Computational Strategy (1)

We divide the system of log-linearised equations in 2 blocks.

Solving the first block

I We take into account all the equations, except the modified UIPcondition.

I We solve this system of equations by the perturbation method, takingthe depreciation rate (∆st) as an exogenous variable.

I The system of log-linear equations become:

A0

[Xt

EtYt+1

]= A1

[Xt−1

Yt

]+A2∆st +B0εt

Back to main .




Solving the second block

I The second block corresponds to the modified UIP condition:

Et∆st+1 = it − i∗t + γσ2($∗t +$∗,cbt ) (1)

I Based on Townsend (1983) and Bacchetta and van Wincoop (2006),we adopt a method of undetermined coefficients conjecturing thefollowing equilibrium equation for ∆st:

∆st = A(L)εi∗t+1 +B(L)ε$

∗t +D(L)ζ ′t (2)

where A(L), B(L) and D(L) are infinite order polynomials in thelag operator L.

Back to main .Marco Ortiz April 2013 22/22



Solving the second block

I We use the solution in the first block to find a MA (∞)representation of the endogenous variables (eg it, $∗cbt ) as afunction of the shocks and replace it in equation (1).

I Signal extraction. Dealers extract information from the observeddepreciation rate (∆st) and signal (∆vd∗t ) to infer the unobservableshocks

(εi

∗t+1, ε

$∗t

):[

∆s∗t∆vd∗t

]=

[a1 b11 0

] [εi

∗t+1

ε$∗

t

]+

[0εvdt

]I Undetermined coefficients: the coefficients in the conjectured

equation (2) need to solve the modified UIP condition (1).

Back to main .Marco Ortiz April 2013 22/22

Documents

FX intervention and monetary policy design: a market … · Show importance of both portfolio-balance/ signaling channels. I FX interventions in a NK-DSGE setup: Benes et al. (2013),