Evolving and relative efficiency of MENA stock markets: evidence from rolling joint variance ratio tests
| Autor | Amira Akl Ahmed |
| Cargo | Lecturer in Economics, Benha University, Egypt |
| Páginas | 91-126 |
Ensayos Revista de Economía–Volumen XXXIII, No.1, mayo 2014, pp. 91-126
Evolving and relative efficiency of MENA stock m arkets:
evidence from rolling joint variance ratio tes ts
Amira Akl Ahmed
Fecha de recepción: 15 I 2013 Fecha de aceptación: 17 I 2014
Abstract
Multiple variance ratio tests, in rolling window procedure, were applied to
weekly data (expressed in local and US dollar currencies) for five stock
markets in the Middle East and North Africa during 1995-2009. Results
indicated that the big and liquid stock markets of Israel and Turkey are
ranked as the most efficient. The Egyptian and Moroccan stock markets
converged towards efficiency by 2002, due to remarkable improvements in
liquidity and information dissemination, wherea s the Jordanian stock markets
restored its efficiency at the end of the st udy period. Exchange rates did not
matter in determining the dynamics of share returns for equity markets
examined.
JEL Classification: G14; G15.
Keywords: Random Walk Hypothesis; Stock Market Efficiency; Variance
Ratio Tests, Wild Bootstrap, Middle East and North Africa.
Resumen
Múltiples pruebas de cociente de varianza, con el procedimiento de
desplazamiento de periodos (rolling window), se aplicaron a datos semanales
(expresados en moneda lo cal y dólares de los E.E.U.U.) para cinco mercados
bursátiles en la región de Medio Oriente y África del Norte ( MOAN), durante
1995-2009. Los resultados indicaron que los mercados bur sátiles de Israel y
Lecturer in Economics, Benha University, Egypt. PhD in Economics, University of
Leicester, England. Address: 18 Atteyah Osman Street, Atreeb, Benha, Egypt. Postcode:
13111. Mobile phone no: 002 011 25999 080. E-mail: amira.ahmed@fcom.bu.edu.eg,
aaa46@alumni.le.ac.uk and amira_akl2002@le.ac.uk
I would like to thank Prof. S. G. Hall (University of Leicester) for his valuable suggestions
which enriched the current research. I highly praise the anonymous referees for their
helpful comments that improved the current work. I am grateful to Edgar Ma ta Flores, a
current PhD student at the department of economics at the University of Leicester, for his
assistance in translating the English abstract into Spanish.
Ensayos Revista de Economía
92
Turquía, grandes y con alta liquidez, están clasificad os como los más
eficientes. Los mercados egipcio s y marroquíes convergieron hacia la
eficiencia en el año 2002, debido a notables mejoras en la difusión de
información y liquidez, mientras que el mercado jordano había restaurado su
eficiencia al final del periodo d e estudio. Los tipos de cambio no participaron
en la determinación de la dinámica de la s ganancias accionarias en los
mercados de capital que fueron examinados.
Clasificación JEL: G14; G15.
Palabras Clave: hipótesis de caminata aleatoria; eficiencia del mercado de
valores; pruebas de cociente de varianza, Bootstrap, Oriente Medio y África
del Norte.
Introduction
According to Fama (1970), a market satisfies the wea k-form efficient market
hypothesis (WFEMH) if relevant information contained in historical prices is
fully, rapidly, and correctly reflected into securities’ prices. The WFEMH is a
joint test of both the fair game property
1
and the validity of the market
equilibrium model incorporated into the hypothesis. The common
equilibrium-pricing model in tests of WFEMH is the hypothesis that expected
returns are constant over time (Fama, 1991). The share price changes because
of fluctuations in expected fundamentals. These fluctuations in expectations
are, in turn, caused by the release of new information that arrives randomly.
Hence, the price o f a share is comprised solely of a permanent (fundamental)
component which is represented by a random walk (RW) model with drift.
Lo and MacKinlay (1988) utilized the pro perty that the variance of the RW is
proportional to time interval and proposed their single variance ratio (VR)
tests. However, their VR tests, which are based on asymptotic
approximations, are biased (severe size distortions and low power) and right-
skewed in finite sa mples, resulting in misleading statistical inference.
Furthermore, it is customary in empirical work to investigate whether the
VRs for several pre-determined holding periods are equal to unit y. The null
of the RW hypothesis (RWH) has to be rejected if it is rejected for some q
1
The WFEMH requires only two necessary conditions. First, it necessitates that the market
is aware of all available and relevant information in the sense that it is not ignored. The
second necessary condition asserts that the market correctly uses the available information
in the sense that the expected return can be viewed as a fair game model in which no
system of trading rules can reap higher expected returns than the equilibrium expected
returns derived by the market. In other words, the actual returns can be randomly greater or
lesser than expected returns, but on average, unexpected returns must be zero ( Ahmed,
2011).
Evolving and relative efficiency of MENA stock markets: evidence...
93
intervals. This sequential procedure results in an oversized testing strategy
since the RWH requires that VRs for all aggregation intervals selected should
equal unity. For this reason, several multiple VR (MVR) tests have been
suggested to overcome these problems. Belaire -Franch and Contreras (2004)
and Kim and Shamsuddin (2008), independently, developed MVR tests based
on the exact rank and sign VR tests of Wright (2000) and the MVR test o f
Chow and Denning (1993). In addition, Kim (2006) introduced a wild
bootstrap version of the MVR test of Chow and Denning (1993).
The current research is motivated by the inconclusive conclusion regarding
testing for WFEMH in five of the Middle East and North Africa (MENA)
countries, na mely Egypt, Jordan, Morocco, Turkey, and Israel [as shown in
table (1)]. Procedures and actions taken by these countries to enhance the role
of the ir stock markets include: (1) relaxation (removal) of restrictio ns
imposed on access of foreign investors to capital markets, (2) adopting
automated trading systems [Egypt (2001), Jordan (2000), Morocco (1997),
Turkey (1993) and Israel (1997)], and (3) regulatory reforms that include
establishment of regulatory bodies to ensure shareholders’ protection and to
monitor market activities (Ahmed, 2013).
From table (1), one could identify t he main reasons behind mixed
conclusions regarding testing for the W FEMH in MENA countries under
consideration. The first reason is using different data frequencies over wide
range of periods. Moreover, traditional tests (e.g. serial co rrelation
coefficients) depend on assumptions that are too restrictive to capture the
patterns in share prices. Another important reason might be employing
different tests with some of s hortcomings (e.g. single VR tests). Furthermore,
with the exception of Yilmaz (1999) and Ahmed (2013), all efficiency studies
mentioned in table (1) tested for the WFEMH in an absolute sense, with an
implicit assumption of efficiency being stead y during the whole study period.
In this context, a more r elevant hypothesis to be tested in the case of
emerging markets and markets under econo mic transition is how such infant
markets converge towards efficiency since it takes time for the price
discovery process to become known (Emerson, Hall and Zalewska-Mitura,
1997). Campbell, Lo and MacKinlay (1997) proposed the concept of relative
efficiency, whic h is the efficiency of one market meas ured agai nst anot her,
indicating that it may be a more useful concept than all-or-none view
investigated by majority of market efficiency literature. Lim and Brooks
(2011) supported using VR methodology in overlapping sub-samples when
testing for WFEMH to capture the gradual change in t he level of efficiency
through time, thereby it would be useful in identifying factors that lead
markets to become (in)efficient. In addition, it may serve as a measure of
constructing efficiency ranking because the main purpose of rolling window
estimation is to measure how frequent the WFEMH is rejected during the
Para continuar leyendo
Solicita tu prueba