Uncovered Interest Parity, Forward Guidance and the Exchange Rate

• DOI: http://doi.org/10.1111/jmcb.12759
• Date: 01 December 2020
• Author: JORDI GALí
• Published date: 01 December 2020

DOI: 10.1111/jmcb.12759

JORDI GALÍ

Uncovered Interest Parity, Forward Guidance and

the Exchange Rate

Under uncovered interest parity (UIP), the size of the effect on the real ex-

change rate of an anticipated change in real interest rate differentials is in-

variant to the horizon at which the change is expected. Empirical evidence

using U.S., euro area and UK data points to a substantial deviation from that

invariance prediction: expectations of interest rate differentials in the near

(distant) future are shown to have much larger (smaller) effects on the real

exchange rate than is implied by UIP. Some possible explanationsare dis-

cussed.

JEL Classications E43, E58, F41

Keywords: forward guidance puzzle, uncoveredinterest rate parity,

unconventional monetary policies, open economy New Keynesianmodel

R    , using a closed

economy New Keynesian model as a reference framework, has emphasized the pow-

erful effects of anticipated changes in interest rates on current output and ination.1

That analysis has drawn a lot of attention given its relevance for the assessment of

the effectiveness of forward guidance policies, that is, the attempt to inuence cur-

rent macroeconomic outcomes in an environment where the zero lower bound (ZLB)

is binding by managing expectations about the future path of the policy rate. On the

other hand, the implausible size of the macroeconomic effects of anticipated changes

in the short term nominal interest rate implied by the standard New Keynesian model

and, in particular, its prediction of a positive relation between the horizon of those

I have beneted from comments by Jón Steinsson, Shogo Sakabe, Philippe Bacchetta, Wenxin Du,

Ken West, Antoine Martin, Marty Eichenbaum, Anna Lipinska, Mick Devereux, Charles Engel, three

anonymous referees, and participants at seminars and conferences at CREI-UPF,Banco de España, SAEe

2017, Sveriges Riksbank, Bocconi, Bank of Canada, ECB, JMCB-NY Fed, NBER IFM Meeting, and

Universityof Edinburgh. I thank Philippe Andrade, Hervé le Bihan, Emanuel Moench and Benson Durham

for help with the data. Angelo Gutiérrez, Christian Hoynck, and Cristina Manea providedexcellent research

assistance. I acknowledge nancial support from the CERCA Programme/Generalitat de Catalunya and

the Severo Ochoa Programme for Centres of Excellence.

J G is a professorat Center for Research in International Economics, Universitat Pompeu Fabra

and Barcelona GSE (E-mail: jgali@crei.cat).

1. See the section below on the background literature for a more detailed discussion and references.

Journal of Money, Credit and Banking, Supplement to Vol. 52, No. S2 (December 2020)

© 2021 The Ohio State University

466 :MONEY,CREDIT AND BANKING

anticipated changes and the size of their immediate effects, has come to be known as

the forward guidance puzzle, and has led many researchers to explore modications

of the assumptions of the standard model in order to get around such implausible

implications.

In the present paper I study, both theoretically and empirically, the effects of an-

ticipated interest rate changes in an open economy, focusing on their impact on the

exchange rate. When doing so, I take the assumption of uncovered interest parity

(UIP, henceforth) and its implications as a theoretical benchmark. This is of particu-

lar interest since most open economy models in the literature generally assume UIP.2

On the other hand, the fact that UIP has been empirically rejected in the literature

does not render the exercise without interest, since it is not clear what are the impli-

cations, if any,of the empirical failure of UIP on the effects of anticipated interest rate

changes on the exchange rate.3The analysis below allows for a comparison between

the predictions of UIP and the empirical evidence with regard to the exchange rate

effects of anticipated variations in interest rates.

As discussed below, UIP makes the current exchange rate depend, to a rst-order

approximation, on the undiscounted sum of expected future interest rate differentials.

Importantly, that relation relies only a relatively weak assumption: the existence at

each point in time of some deep pocket investors with unconstrained access to both

domestic and foreign bonds.

In the rst part of the paper I analyze the effects of anticipated interest rate differ-

entials on the exchange rate, under the assumption of constant prices (i.e., ignoring

any induced effects on ination). In that environment, the combination of UIP with

the long run neutrality of monetary policy yields a strong implication: the impact on

the current exchange rate of an anticipated future adjustment of the nominal rate is

invariant to the timing of that adjustment.

Next I revisit the analysis of the effect of anticipated changes in interest rate dif-

ferentials while allowing for feedback effects on output and prices, using a simple

New Keynesian model of a small open economy. I show how, in that environment,

the effect of a given anticipated change in the short-term nominal rate on the current

exchange rate is larger the longer is the horizon of implementation. A similar predic-

tion applies to the effect on output and ination. As discussed below,both results are

closely connected to the so called forward guidance puzzle uncovered in the recent

literature, though that literature has invariably focused on closed economy models

and has thus ignored the real exchange rate channel.4

Having analyzed the theoretical implications of UIP on the link between antici-

pated interest rates and the exchange rate, in the second part of the paper I turn to the

2. The present paper is not, by any means, the rst to study, from a theoretical point of view, the

impact of anticipated monetary policies on the exchange rate and other variables. Early work addressing

that question includes Wilson(1979), Dornbusch and Fischer (1980) and Mussa (1982). In contrast with the

present paper, however,that work, made use of not-fully-microfounded models. Most importantly, it did

not examine the role played by the horizon of anticipated policies, which is the focus of the present paper.

3. See, for example, Bacchetta (2013) and Engel (2014) for a survey of the empirical literature on UIP.

4. See Carlstrom et al. (2015), Del Negro et al. (2015), and McKay et al. (2016, 2017), among others.

JORDI GALí :467

data. In particular, I provide some empirical evidence on the exchange rate effectsof

anticipated future interest rate differentials at different horizons. An open question,

which is the focus of the present inquiry, is what the empirical failure of UIP im-

plies with regard to the link between the current real exchange rate and anticipated

real interest rate differentials at different horizons. Thus, a key objective of the em-

pirical analysis below is to characterize the potential empirical deviations from the

horizon-invariance property implied by the UIP.5

Using data for the U.S., UK and euro area on bilateral real exchange rates and

market-based proxies for anticipated real interest rate differentials at different hori-

zons, I test the horizon-invariance property linking those variables, as implied by the

UIP.The evidence points to a strong rejection of that property. Perhaps more interest-

ingly, it suggests a simple characterization of the empirical deviations from horizon-

invariance:expectations of interest rate differentials in the near (distant) future appear

to have much larger (smaller) effects than predicted under UIP. As far as I know,this

particular dimension of the empirical failure of UIP, which I refer to as the forward

guidance exchange rate puzzle, has not been uncovered in the existing literature.

The third part of the paper discusses possible interpretations of the empirical nd-

ings. In particular, I argue that some of the solutions to the forward guidance puzzle

proposed in the closed economy literature are unlikely to apply to the exchange rate

channel emphasized in the present paper. Instead I propose a simple behavioralmodel

that is shown to be consistent with the key qualitative ndings uncovered in the em-

pirical section.

The remainder of the paper is organized a follows. Section 1 briey describes the

forward guidance puzzle in a closed economy setting. Section 2 discusses the ef-

fects of forward guidance on the exchange rate in a partial equilibrium framework.

Section 3 revisits that analysis in general equilibrium, using a small open economy

New Keynesian model as a reference framework. Section 4 presents the empirical

evidence. Section 5 discusses possible interpretations of the evidence. Section 6 sum-

marizes and concludes.

1. BACKGROUND: THE FORWARD GUIDANCE PUZZLE IN THE CLOSED

ECONOMY

In this section, I briey review the literature on the forward guidance puzzle.

The analysis in that literature has been invariably conducted using a closed econ-

omy framework.

The effectiveness of forward guidance and its role in the design of the optimal

monetary policy under a binding ZLB was analyzed in Eggertsson and Woodford

(2003) and Jung et al. (2005), using a standard New Keynesian model. Those papers

emphasized the high effectiveness of forward guidance as a stabilizing instrument, as

5. Severalrecent papers have analyzed the response of the exchange rate to news about future monetary

policy, butwith a different focus from the one adopted here. See below for discussion and references.