Exchange rate behaviour with negative interest rates: Some early negative observations

• Author: Andrew K. Rose,Allaudeen Hameed
• DOI: http://doi.org/10.1111/1468-0106.12250
• Date: 01 February 2018
• Published date: 01 February 2018

SPECIAL ISSUE ARTICLE

Exchange rate behaviour with negative interest

rates: Some early negative observations

Allaudeen Hameed

1

| Andrew K. Rose

2†

1

NUS Business School, National University of

Singapore, Singapore

2

Haas School of Business, University of

California, Berkeley, California

Correspondence

Andrew Rose, Haas School of Business,

University of California, Berkeley CA 94720-

1900.

Email: arose@haas.berkeley.edu

†

The data set, key output, and both a longer

version and current copy of the paper are freely

available at Rose’s website.

Abstract

This paper examines exchange rate behaviour during the

recent period with negative nominal interest rates. We use

a daily panel of data of 61 currencies from January 2010

to May 2016; during this time five economies (Denmark,

EMU, Japan, Sweden and Switzerland) experienced nega-

tive nominal interest rates. We examine both effective

exchange rates and bilateral rates, the latter typically mea-

sured against the Swiss franc because Switzerland has had

the longest period of negative nominal rates. We examine

exchange rate volatility, exchange rate changes, devia-

tions from uncovered interest parity, and profits from the

carry trade. We find that negative interest rates seem to

have little effect on observable exchange rate behaviour.

1|INTRODUCTION

This paper examines exchange rate behaviour during the recent period of negative nominal interest

rates. We use a panel of daily data between January 2010 and May 2016 for 61 economies to exam-

ine exchange rate volatility and deviations from uncovered interest parity.

1

During this time, five

economies (Denmark, EMU, Japan, Sweden and Switzerland) experienced nominal interest rates that

were (non-trivially) negative.

2

Our results are mostly negative; we find little evidence that negative

interest rates have had any substantial effect on exchange rates.

2|LITERATURE

There is a small but growing literature on the effects of negative nominal interest rates. References

of relevance include Viñals, Gray, and Eckhold (2016) and Arteta, Ayhan Kose, Stocker, and Taskin

(2016). The key questions addressed about negative nominal rates usually concern their efficacy and

1

We often refer to these informally below as countries, even though currency unions like the Euro zone include a number of

countries.

2

At least in the relevant sense. In late March 2016, Hungary lowered its overnight deposit rate to −0.05%, although the more relevant

base rate remained positive and much higher (1.2%).

Received: 1 December 2016 Revised: 9 February 2017 Accepted: 1 April 2017

DOI: 10.1111/1468-0106.12250

© 2018 Asian Development Bank Institute (ADBI)

Pacific Economic Review © 2018 John Wiley & Sons Australia, Ltd

Pac Econ Rev. 2018;23:27–42. wileyonlinelibrary.com/journal/paer 27

limitations in stimulating economic growth and inflation, and their effect on bank profitability and

financial stability. Little attention is usually paid to the exchange rate, although it is widely recog-

nized that negative nominal interest rates were sometimes introduced to deter capital inflows, as in

the cases of Denmark and Switzerland.

The literature gives us little reason to believe that the introductionor maintenance of negative nom-

inal interest rates has a material effect on exchange rate behaviour. For instance, Arteta et al. (2016,

pp. 26–46) write “The currencies of NIRP [negative interest rate policy] countries have shown varied

responses …Enduring changes in exchange rates and equity market indexes cannot be discerned from other

factors affecting them over time …the exchange rate response [of emerging and developing economies] var-

ied considerably across countries, both in terms of size and direction …,”while Viñals et al. (2016, p. 2)

write “The impact of negative central bank rates on the exchange rate has been mixed ….”

Still, there is reason to look for a discontinuity in exchange rate behaviour around a nominal

interest rate of zero. Perhaps the shocks that drive nominal interest rates negative are different from

those when rates are positive but low. Perhaps foreign exchange market participants perceive there

to be some discontinuous effect because of subtle changes to arbitrage conditions like covered inter-

est parity. Accordingly, we conduct an open-minded empirical exploration.

3|THE DATA SET

We are interested in understanding exchange rate behaviour during the contemporary period of neg-

ative nominal interest rates. Negative interest rates are a recent phenomenon, so we wish to maxi-

mize the potential scope of a necessarily limited data set. We begin the data set in January 2010, to

reduce the after-effects of the Global Financial Crisis and Great Recession while also including a

period of comparable data before the onset of negative interest rates. We rely on the highest fre-

quency of data (daily) that is reliably available for a wide range of countries because we hope to

bring a large cross-section to bear on a problem with a necessarily limited time-series span. We treat

Switzerland as the base country for much of the analysis because Switzerland was the first economy

to hit negative interest rates recently.

3

During the sample period, Switzerland experienced 827 days

of negative nominal interest rates; it has also experienced the largest negative interest rates in abso-

lute value.

4

Accordingly, we convert bilateral dollar (and pound) rates to Swiss rates (foreign cur-

rency per Swiss franc), assuming trilateral arbitrage. However, for a sensitivity analysis we also use

the US dollar, the pound sterling and the euro as alternative bases.

Our spot exchange rates are closing rates calculated by the WM Company based on data pro-

vided by Reuters at or around 16:00 hr London time.

5

These rates are determined close to the mid-

dle of the “global day”(11:00 hr New York time) during a time of high liquidity in the global

foreign exchange market. We primarily use mid-point bilateral US dollar rates as primitive data but

3

Again, Sweden technically had negative interest rates from July 2, 2009 to September 2, 2010, because the (1-week) repo rate was

set to 0.25% and the 1-week deposit rate was mechanically cut at that point to −0.25%. However, given the small size of the relevant

market, this appears to have been a technicality (http://economix.blogs.nytimes.com/2009/10/01/negative-interest-rates-in-sweden/).

Sweden ended this period on September 2, 2010.

4

During the sample, Switzerland experienced more than twice as many days of negative nominal interest rates as EMU, Denmark and

Sweden; Japan has far fewer still.

5

A number of snapshots are taken from the Reuters system around 16:00 h and median rates are then selected for each currency. This

is done independently for bid and offer quotes. When the rates have been validated WM derive cross rates to GBP and EUR (or GBP

and USD). Mid rates are calculated as the arithmetic mean of bid and offer. WM/Reuters monitor national holidays in the USA, the

UK, Germany and Japan, and if two or more of these are open a fixing is produced (if only one is open, generally rates from the pre-

vious weekday are used; no fixing is produced on December 25 or January 1).

28 HAMEED, AND ROSE