Long memory or structural breaks: Some evidence for African stock markets

• Author: Geoffrey Ngene,Ali F. Darrat,Kenneth A. Tah
• Published date: 01 September 2017
• Date: 01 September 2017
• DOI: http://doi.org/10.1016/j.rfe.2017.06.003

Long memory or structural breaks: Some evidence for African

stock markets

Geoffrey Ngene

a,

⁎, Kenneth A. Tah

a

, Ali F. Darrat

b

a

StetsonSchool of Business and Economics,Mercer University,Macon, GA 31207, United States

b

Collegeof Business, LouisianaTech University, Ruston, LA 71272,United States

abstractarticle info

Articlehistory:

Received20 May 2016

Receivedin revised form 29 May 2017

Accepted12 June 2017

Availableonline 16 June 2017

JEL classification:

G12

G14

G15

This studyexplores two main issues.One: How do the exchange rate and inflationrate impact long termdepen-

dencystructure of weekly returnsof seven African stockmarkets? Two: What is the long termdependency struc-

ture of stockreturns and variance in the absenceand in the presence of structural breaks?The sample period is

06-2002through 02-2014.The returns are de-thinnedto accountfor illiquidity of stockreturns. The studyimple-

ments sequential multiplestructural breaks test to identify structural breaks. Threesemi-parametric tests and

three parametric methods are th en implemented to test for the long memory behavior of raw returns and

break-adjustedreturns. The longmemory estimates decline monotonically in magnitudeor in statistical signifi-

cance as returnsare adjusted for exchange rate andthen inflation rate. There is evidence of multiplestructural

shifts whose breakdates correspond to major domestic or globalevents. Without structural breaks,the results

indicate the existenceof long memory components in returns and variance. Using the break-adjustedreturns,

the long memory evidencein returns and variance weakens in magnitudeand/or statistical significanceor just

dissipatesin some ASMs. The absence of longmemory using break-adjustedreturns suggests thatlong memory

may be an artifact of short memory encounteringbreaks in some ASMs. Caution is therefore warranted when

interpretinglong memory inferencesin the presence of structural breaksand making policy recommendations.

Our evidenceis useful to investors, fund managersand policy makers.

© 2017 Elsevier Inc. All rights reserved.

Keywords:

Long memory

Marketefficiency

Stockreturns

Africanmarkets

Structuralbreaks

1. Introduction

In its semi-strongform, the stock marketefficiency (SME) hypothe-

sis contendsthat stock prices fully incorporate any new information as

soon as it becomespublicly available.Therefore, there shouldbe no ex-

ploitable components in stock prices for predicting future prices and

earning abnormal returnswithout additional risk. Long memory—also

called long-range dependencyor persistency—occurs when the statisti-

cal dependencein stock prices returnsdecays slowly overtime. Accord-

ing to Holmes and Grimes (2008), the degree of persis tence of asset

prices is a key determinantof macroeconomic stability of any country.

The degree of persistence also helps in formul ation and adoption of

macro policiesafter an exogenous shock.For example: In caseof a pos-

itive shock to an African stock Market (ASM) stock price series which

exhibitsstationary long memory (meanreversion), strong policyinter-

ventionis required to maintainthe series at the new trajectoryinstigat-

ed by the positive shock. Likewise, a negative shock affecting a series

characterized by non-stationary long m emory (unit-root) requires a

strong policyintervention to avoidformation of bubbles and toredirect

the series to its longterm equilibrium path or value.

Long memory implies that current assetchanges in asset prices are

heavilydependent on distant pastprice changes. This makesit possible

for future pricechanges to be predicted from past pricechanges. If, in-

deed, investorshave a priori knowledge regardingthe persistent stock

prices changes, they can reap huge profits by buying (selling) st ocks

when prices are expected to rise (fall) from the mere observation of

the persistence of price behaviors. This would not onlycontradict the

martingaleprocess assumed in most financial assetpricing models but

the presenceof long memory processwould also contravenemarket ef-

ficiency.Market inefficiency,which is characterizedby mispricedassets,

may leadto inefficient allocationof capital among householdsand insti-

tutionalinvestors. In fact, Lo(1991) suggests that long memoryin asset

returns makesportfolio allocation andconsumption (saving) decisions

sensitiveto investment horizons.

Bekaert and Harvey(1998) also argue that informational efficiency

affords a fundamental linkage between stock marke ts and economic

growth inemerging economies.Therefore, the presenceof long memo-

ry and accompanyingmarket inefficiencymakes it considerablyimpor-

tantfor policy makers andregulators of Africanstock markets(ASMs) as

they formulate and implement poli cies that mitigate market

Reviewof Financial Economics 34 (2017)61–73

⁎Correspondingauthor.

E-mailaddress: ngene_gm@mercer.edu(G. Ngene).

http://dx.doi.org/10.1016/j.rfe.2017.06.003

1058-3300/©2017 Elsevier Inc. All rightsreserved.

Contents listsavailable at ScienceDirect

Review of Financial Economics

journal homepage: www.elsevier.com/locate/rfe