Do Monetary Policy Announcements Affect Exchange Rate Returns and Volatility of Returns? Some Evidence from High‐Frequency Intra‐Day South African Data

• Date: 01 September 2018
• Author: Cyril May,Jannie Rossouw,Greg Farrell
• Published date: 01 September 2018
• DOI: http://doi.org/10.1111/saje.12196

DO MONETARY POLICY ANNOUNCEMENTS AFFECT

EXCHANGE RATE RETURNS AND VOLATILITY OF

RETURNS? SOME EVIDENCE FROM HIGH-FREQUENCY

INTRA-DAY SOUTH AFRICAN DATA

CYRIL MAY*, GREG FARRELL

†,‡

AND JANNIE ROSSOUW

†

Abstract

This paper examines the temporal effect of domestic monetary policy surprises on both returns and

volatility of returns of the South African rand/U.S. dollar exchange rate. The analysis in this “event

study” proceeds using intra-day minute-by-minute exchange rate data, repo rate data from the South

African Reserve Bank’s scheduled monetary policy announcements, and market consensus repo rate

forecasts. A carefully selected sample over the period August 2003 to November 2017 ensures that

the change in monetary policy is exogenous to the exchange rate. We find statistically and economi-

cally significant responses in intra-day high-frequency South African rand/U.S. dollar exchange rate

returns and volatility of exchange rate returns to domestic interest rate surprises, but anticipated

changes have no bearing on exchange rate returns and their volatility. The empirical results also indi-

cate that there is an instantaneous response of the rand/dollar exchange rate to monetary policy sur-

prises and that monetary policy news is an important determin ant of the exchange rate until at least

42 minutes after the pronouncement – suggesting a high degree of market “efficiency” in its mechan-

ical sense (although not necessarily in the deeper economic-informational sense) in processing this

information. Essentially, the asymmetric GARCH results exhibit no leverage effects – positive and

negative information shocks have symmetric effects on conditional variance.

JEL Classification: C22, E52, E58, F31, F41, G14, G15

Keywords: Exchange rate, expectations, monetary policy surprises, repo rate, returns, volatility

1. INTRODUCTION

In this study, we examine the behaviour of the South African rand/ U.S. dollar exchange

rate (hereafter referred to as the rand/dollar exchange rate) returns and conditional var-

iance (or volatility) of returns in reaction to domestic monetary policy announcements,

employing an event study approach and using intra-day high-frequency (1-minute slice)

exchange rate data from 2003 to 2017. More specifically, the paper examines how rand/

dollar exchange rate returns and their volatility digest information contained in the sur-

prise component of scheduled repo rate announcements – how soon exchange rate

returns and return volatility respond after this news release, to what extent exchange rate

* Corresponding author: Lecturer and Researcher, School of Economic and Business Sciences,

University of the Witwatersrand, Johannesburg, South Africa. Tel: 127117178104, Fax:

1277178081. E-mail: cyril.may@wits.ac.za

†

School of Economic and Business Sciences, University of the Witwatersrand, Johannesburg,

South Africa

‡

South African Reserve Bank, Pretoria, South Africa

V

C2018 Economic Society of South Africa. doi: 10.1111/saje.12196

308

South African Journal of Economics Vol. 86:3 Sept ember 2018

South African Journal

of Economics

returns and their volatility react, and how long the news effect lasts. The “surprise” or

unexpected component of the repo rate announcement is defined here as the difference

between the actual rate announcement and the market consensus median rate forecast.

Analysing the response of nominal exchange rate returns and the volatility of exchange

rate returns to economic and noneconomic news, in developed and emerging markets,

has become a very active research area in international finance over the past decade or so.

This study is motivated by several factors. The use of an event study methodology – an

assessment of the impact of an event on the price of a financial asset around the time of

the event shortly before and/or after the event(s) – is useful for examining how central

bank communications affect financial markets (Haldane, 2017). From an econometric

perspective, short window periods enable us to target the effect of specific announce-

ments over a precisely defined time interval, abstracting from the joint determination of

interest rates and exchange rates by confidently isolating events in which causality runs in

only one direction, from interest rates to exchange rates (G€urkaynak and Wright, 2013).

Furthermore, policy announcement reactions provide a unique window into how the

economy operates because the efficient market hypothesis (EMH) implies that asset pri-

ces should react directly and quickly to the surprise component of any announcement.

Therefore, this study provides an opportunity to make a meaningful contribution to the

existing South African literature.

This paper contributes to the South African literature on exchange rate responses to

monetary policy repo rate announcements in three ways. This is the first comprehensive

study on South African interest rate announcement effects using intra-day high-frequency

(minute-by-minute) exchange rate data; Fedderke and Flamand (2005) and Mpofu and

Peters (2017) employ daily exchange rate data. Second, in addition to estimating the

exchange rate returns reaction function, exchange rate return volatility responses over a

protracted period are also considered; Maserumule and Alagidede’s (2017) 10-minute

window volatility analysis focuses on a shorter 2-year span sample.

1

And third, this study

covers a much longer sample of high-frequency data than has previously been used in

South African studies, allowing for better estimates and a more thorough examination of

responses to be made. In particular, our sample covers most of the inflation-targeting

regime – 2003 to 2017 – allowing more time for the South African Reserve Bank’s

(SARB’s) inflation-targeting framework to become entrenched. Although the inflation-

targeting policy was adopted by South Africa in 2000, the first target was for the calendar

year 2002. The exact sample period in this study, August 2003 to November 2017, is

dictated by the availability of historical market consensus forecasts for the repo rate,

adequate information regarding the monetary policy committee (MPC) repo rate deci-

sion and the intra-day high-frequency exchange rate data. This paper significantly extends

the sample periods for earlier similar studies on the rand/dollar exchange rate: (i) Fed-

derke and Flamand’s (2005) sample period, June 2001 to June 2004, covers the episode

when the rand plummeted from a low of 13.85 against the dollar in 2001 to a strong

recovery to around 6 to the dollar in 2004; (ii) Mpofu and Peters (2017) examine data

for the period March 2000 to December 2014; and (iii) Maserumule and Ala gidede’s

(2017) dataset covers the sample period January 2014 to December 2015.

1

Farrell et al. (2012) also use high-frequency data but look at South African inflation and not

interest rate surprises. They also do not consider the impact of announcements on volatility.

V

C2018 Economic Society of South Africa.

309

South African Journal of Economics Vol. 86:3 Sept ember 2018