Cointegration and Causality between Financial Development and Economic Growth: Evidence from Morocco
Searching for Determinants of Pay or Not to Pay Cash Dividend in Indonesia
Triasesiarta Nur and Kresnohadi Ariyoto Karnen
Month of the Year and Pre-Holiday Effects in Indonesia and Malaysia Shari’ah Compliance
Helma Malini and Mohamad Jais
The Impact of Pre-Closing Implementation to Price Efficiency in Indonesia Stock Exchange
Gilang Praditiyo
The Determinant Factors of Real Estate Investment Trust (REIT)’s Performance: Evidence from Asian REITs
Nor Edi Azhar Binti Mohamad and Ilyas Ariefin Bin Zolkifli
Management Research Center (MRC)
Department of Management, Faculty of Economics Universitas Indonesia Depok Campus, West Java, Indonesia Phone : +62 21 7272425 ext. 503 Fax
Vol. VI
Issue 1
Pages
Depok ISSN
UNIVERSITAS INDONESIA Faculty of Economics and Business Department of Management
Editor in Chief
Ruslan Prijadi, Universitas Indonesia
About MRC
independently or in collaboration with other research organization and also with private sectors. The field
Vice Editor
of these activities from informal discussion group Rofikoh Rokhim, Universitas Indonesia
The Management Research Center (MRC)
was created in March 2006 to strengthen quality
to major international events. Especially for the public lecturer, MRC invites
and opens the opportunity for distinguish professor, Ito Warsito, President Director of Indonesia Stock Exchange
Institutional Board of Editor
of management research in Indonesia. MRC is
Nurhaida, Indonesia Capital Market and Financial Institution Supervisory Agency
national research center dedicated to contribute
the theoretical, empirical, and practical research in
leaders and CEO of the companies to share their
know ledges and their experiences. MRC also holds
recent management issues.
Managing Editor
three annual events: International Conference on Dwi Nastiti Danarsari
The intention is not only to stimulate research
and discussion within scholarly circles, but also to
Business and Management Research (ICBMR), Doctoral Journey in Management (DJM), and
enhance business community and public awareness to stimulate thinking on and exploring solutions in
Layout and Typesetting
Master Journey in Management (MJM). Shafruddin Nusantara
management issues.
Administration
The MRC is placed to assist local, regional
Publications
Angtyasti Jiwasiddi
and international scholar and other researcher that provides a congenial and stimulating intellectual
To facilitate timely dissemination of research and
environment, encouraging the fullest interaction
commentaries on recent and current developments,
EDIToRIAL ADVISoRY BoARD
and exchange ideas.
MRC produce monographs under the Working
To achieve these aims, the MRC conducts a
Eliseo A. Aurellado,
Eduardus Tandelilin,
Paper Series.
Ateneo De Manila University
Universitas Gadjah Mada
range of research programs; holds public lecturers,
MRC also publishes academic journals namely,
seminars, workshops, and conferences; publishes
The South East Asian Journal of Management
Henk von Eije,
(SEAM), Indonesian Capital Market Review University of Groningen
Irwan Adi Ekaputra,
research journals and books, support research
Universitas Indonesia
facilities, including data, financial subsidy and
(ICMR), and ASEAN Marketing Journal (AMJ).
library collections.
Jean-Pierre Laffargue,
Sugeng Wahyudi,
Universite of Paris I Pantheon - Sorbonne
Universitas Diponegoro
Library
Research
Ralf-Yves Zurbrugg,
The MRC library collaborates with the Graduate University of Adelaide
Research within MRC covers the fields of interest
School of Management, Faculty of Economics,
Universitas Indonesia. This library has many Universitas Indonesia
in:
Bambang Hermanto,
• Business Policy & Strategy
excellent collections of management studies
• Entrepreneurship
in Bahasa Indonesia and English, in print and
• Finance and Banking
multimedia formats.
• Capital Market
The Indonesian Capital Market Review
(ISSN 1979-8997) is dedicated to exploring, disseminating and discussing the
• Gender and Diversity in Organizations
latest issues and developments of capital markets, particularly those related or relevant to the advancement of capital
Contact us
• International Management
market in Indonesia. The mission of the editorial board of ICMR is to present the latest thinking and research that tests,
extends, and develops management theories or those contributes to capital market practices.
• Management Education & Development • Operations Management Management Research Center (MRC)
• Organization Development & Change
Co-published twice a year (January and July) by:
• Organization & Management Theory Department of Management, Faculty of Department of Management, Faculty of Economics,Universitas Indonesia
Management Research Center (MRC)
• Organizational Behavior Economics, Universitas Indonesia
Depok Campus, West Java, Indonesia
Indonesia Stock Exchange (IDX)
• Research Methods in Management • Social Issues in Management
Indonesian Corporate Secretary Association (ICSA)
Phone : +62 21 7272425 ext. 503
• Technology & Innovation Management
Management Research Center (MRC),
Email : mrc@ui.ac.id
Department of Management, Faculty of Economics, Universitas Indonesia
Activities
Depok Campus, West Java, Indonesia Phone : +62 21 7272425 ext. 503 Fax
MRC has a program of public lecturers,
seminars, workshops, and conferences, organized
Email : icmr@ui.ac.id and icmr.feui@gmail.com
Copyright © 2014 Management Research Center (MRC), Department of Management, Faculty of Economics,
January 2014 • VOL.VI • NO.1
Contents
Cointegration and Causality between Financial Development and Economic Growth: Evidence from Morocco
Abdellatif Chatri and Abdelouahab Maaruf ____________________________________
Searching for Determinants of Pay or Not to Pay Cash Dividend in Indonesia
Triasesiarta Nur and Kresnohadi Ariyoto Karnen _______________________________
Month of the Year and Pre-Holiday Effects In Indonesia and Malaysia Shari’ah Compliance
Helma Malini and Mohamad Jais ____________________________________________
The Impact of Pre-Closing Implementation to Price Efficiency in Indonesia Stock Exchange
Gilang Praditiyo _________________________________________________________
The Determinant Factors of Real Estate Investment Trust (REIT)’s Performance: Evidence from Asian REITs
Nor Edi Azhar Binti Mohamad and Ilyas Ariefin Bin Zolkifli _______________________
Cointegration and Causality between Financial Development and Economic Growth: Evidence from Morocco
Abdellatif Chatri* and Abdelouahab Maaruf**
Mohamed V University, Rabat, Morocco
The debate on the relationship between economic growth and financial development has been steadily growing in these recent years. However, the existing theoretical and empirical literature pro- vides conflicting views in this respect. This paper proposes an empirical investigation of the nature of this relationship in the Moroccan context. More precisely, it explores the cointegrating and the cau- sality issue between economic growth and financial development. The latter is measured by largely used indicators. In particular, we use capital market proxy, in addition to the traditional indicators of financial intermediation. The findings show that financial development explains significantly the growth, but the direction of causality depends on the indicator used to measure the financial deepen- ing and the time horizon of analysis (short or long terms).
Keywords: Financial development, Economic growth, Cointegration, Causality, VECM
Introduction
long-time conventional ideas have been tested on this occasion.
The international financial crisis of 2008 is Indeed, it is nowadays recognized that the in-
a seizing phenomenon in several respects. It stability due to the financial innovation process is striking by its scope, its rapid contagion to
development can likely penalize in return the the global financial system, but also according
economic growth and annihilate, consequently, to the intervention of public authorities, by the
the positive effects of financial development. unprecedented measures and new modalities,
The dramatic decline of the potential growth of in order to break the spiral of distrust in the
OECD countries by nearly three points (OCDE financial system. Moreover, it is appealing by
(2010) p.243), under the influence of the last its cost and its duration, as five years later, its
financial crisis is the eloquent proof. effects and repercussions on the real economy
The positive impact of financial develop- continue to be drastically felt.
ment on economic growth (McKinnon, 1973; Throughout this crisis, which is one of the
Shaw, 1973) is not a systematic. It is true that most impressive and undeniably unparalleled
most studies conclude that there is a positive since World War II, there has been an abun-
relationship between these two sectors (King dance of researches on the nexus between the
and Levine,1993, p.718; Levine, Loayza and financial sector and the real economy 1 . Various Beck, 2000, p.31; Benhabib and Spiegel, 2000,
*Faculty of Law and Economics,Mohamed V University Avenue des Nations-Unies, B.P. 721 Agdal - Rabat – Maroc, E-mail: chatriabdellatif@yahoo.fr. **Faculty of Law and Economics,Mohamed V University Avenue des Nations-Unies, B.P. 721 Agdal - Rabat – Maroc, E-mail: abdelmaarouf@gmail.com.
1 The analysis of Bagehot ( 1873 ) and Schumpeter ( 1911 ) are often considered as precursors in this respect.
INDONESIAN CAPITAL MARKET REVIEW • VOL.VI • NO.1
p.341; Christopoulos and Tsionas, 2004, p.55; similar works on Moroccan case (Solhi, 2006; Giuliano and Ruiz-Arranz, 2009, p.144). How-
Abouch et Ezzahid, 2006; Alaoui, 2004). From ever, some studies show that this relationship
our point of view, investigating the financial can be negative (Kaminsky and Reinhart, 1999,
development and economic growth relation- p.494; Demirgüc-kunt and Detragiache,1999,
ship makes sense, only when the direct regula- p.35), and a non-lineardue to the presence of
tions are removed. That is why we have taken threshold effect (Greenwood and Jovanovich,
as a starting point of our analysis the year 1998, 1990, p.1076; Berthélemy and Varoudakis,
which marks the achievement of the financial 1996, p.11; Deidda and Fattouh, 2002, p.339),
liberalization process in Morocco and coin- or also inexistent (Robinson (1952), Lucas
cides also with the new base year of national (1988)). Besides, some studies have explored
accounts.
the interaction between those sectors, with even The last but not the least contribution of this there ambivalent results (Agbetsiafa, 2003;
paper lies in the fact that it goes above the study Waqabaca, 2004; Odhiambo, 2004; Fowowe,
of the link between financial and economic sec- 2010).
tors, but aims to approach the question of their Considering these diverse and contradictory
causality.
proposals, this paper aims to provide an empiri- The rest of the paper is organized as follows. cal evidence on the relation between financial
The first section presents a brief review of the development and economic growth in the Mo-
literature. The second one exposes our model. roccan context.The importance of this paper
The third presents and discusses the results. stems from the following elements. Firstly, it concerns the particular case of Mo-
Literature Review
rocco, where the financial sector is considered as a model on the regional area.Thus, the re-
The arguments supporting the importance forms undertaken to deepen and modernize the
of the financial system in stimulating economic financial sector, were not only worth wiping out
growth are not new. Schumpeter (1911) high- successfully the international financial crisis th lighted at the beginning of the 20 century the
(Chatri, Maarouf et Zouiri, 2013, p.178), but role of banks in the financing innovations. Dur- also leading to have a financial sector enough
ing the 50’s, Gurley and Shaw (1955) stressed diversified and endowed with all the attributes
the importance of financial intermediation in of a modern financial sector (IMF, 2008, p.5).
economic growth. Inspired by these ideas, Pat- Consequently, the case of Morocco is an ex-
rick (1966) shows that efficient financial system cellent example to approach the nature and the
increase saving and its allocation to productive extent of the relationship between financial de-
investments.
velopment and economic growth. It was not until the early 70’s when the link Secondly, if the majority of the similar works
"finance- economic growth "started to be treat- have examined this relationship within the con-
ed in a detail, through the school of financial text of the panel data (Yoke-Kee et Muzafar,
repression developed by Goldsmith (1973), 2011; Rachdi et Ben Mbarek, 2011; Abu-Bader
McKinnon (1973) and Shaw (1973). Accord- et Abu-Qarn, 2008; Ben Naceur et Ghazouani,
ing to these authors, the reduction by monetary 2005; Bernard, 2000; Turunc, 1999), the pre-
authorities of the nominal interest rates discour- sent work focuses deliberately on the specific
ages savings and fixes investment below the and isolated case of Morocco. This permits to
equilibrium level that would be necessary for avoid biased results related to specific politi-
the economic growth. They suggest liberalizing cal, institutional and economic factors of each
the financial sector in order to increase savings country.
and investment and improve capital efficiency. Thirdly, it uses the homogeneous data with-
With the development of the endogenous out apparent structural bias, contrary to several 2 growth theory in the 80’s the debate on the
2 In order to overcome inadequacies of the neoclassical growth model which cannot explain the mechanisms those gener-
Chatri and Maaruf
link between financial development and growth
3) has known a considerable revival. Within this framework, the models developed show that
I t = ϕS t
Taking into account (1), the growth rate in financial development is an important channel
year t +1 can be expressed as: through which growth in the stationary state oc-
curs endogenously. 4) In this regard, one of the first models which
established a theoretical framework of positive Equations (2) and (3) allow to deduce the relationship between financial intermediation
steady growth rate g:
and economic growth is developed by Ben- civenga and Smith (1991). Based on the theo- retical analysis of financial intermediation of Gurley Show, the aforementioned authors have developed an overlapping-generations mod-
5) el in which the banking system plays an im-
portant role. The latter provides funds for long where s = S / Y denotes the gross saving rate. term productive investment and allows, by pro-
The equation (5) shows that financial devel- viding liquidity, investors to hold bank deposits
opment can affect growth through three chan- rather than liquid and unproductive assets 3 .
nels: i) increasing the proportion of savings (ø) This article is deepened by several other
allocated to investment productive, ii) improv- works (De Gregorio, 1992; Roubini-I-Martin,
ing the marginal productivity of capital (A) iii) 1992; Saint-Paul, 1992; King et Levine, 1993),
and increasing savings rate of the economy (s). each of which focused on a very particular as-
In parallel to this theoretical framework, a pect of finance-growth relationship. A more
rich empirical literature demonstrating how complete analysis was presented by Pagano
inefficient financial system can positively in- (1993), who has identified three main channels
fluence the long term growth rate has been through which financial system can influence
emerged (among others Greenwood and Jo- growth, starting from the following model de-
vanovic (1990), Levine (1991), Bencivenga veloped by Rebelo (1991):
and Smith (1991). The empirical work of King Y = AK
and Levine (1993) is highly important to men-
tt
tion here. Its importance lies in the fact that it where Y, A, K are respectively the produc-
confirms the existence of a strong relationship tion, the marginal productivity of capital and
between growth and financial development for the capital stock. He introduced for this purpose
a broad sample composed of 80 countries. In the conventional model of the gross investment
addition, it is based on various relevant indica- (I t ):
tors of financial development on growth. =K t+1 -
Levine (1991), Bencivenga, Smith and Starr
I t (1-δ)K t
(1996) have based their works on financial where the coefficient (δ) denotes the capital
market indicators. They demonstrate that the depreciation. It assumes also that the fraction
liquidity of stock market is crucial as it facili- (1-ø) of capital is absorbed by financial inter-
tates investment in long term projects and thus mediation (cost of intermediation and pruden-
stimulates economic growth. In the same line, tial rules):
Greenwood and Smith (1997) demonstrate that
ate the growth in stationary state.According to this model, developed by Solow (1956), economic growth is explained by two exogenous variables: demographic factors and technology. For more details on the theories of growth, see for example Aghion, P. and Howitt, P (2010), "l’économie de la croissance" Economica.
3 We find here the findings of Diamond and Dybvig who have shown that banks are relevant according to their "endog- enous response" to the divergence on time horizons consumption. Cf, Diamond, D.W. Dybvig, P.H. « Bank runs, deposit
insurance and liquidity », Journal of Political Economy, Vol. 91, No. 3, 1983, pp. 401-419.
INDONESIAN CAPITAL MARKET REVIEW • VOL.VI • NO.1
a developed stock market can reduce the cost inson (1952) considers that financial sector is of mobilizing savings and facilitate productive
not relevant on economic development. Lucas investment. Taking into account the global mar-
(1989) suggests also that the role of financial ket, Rousseau and Wachtel (2000) have also
factors in economic growth is exaggerated. stressed the benefits of deep and liquid financial
In the same line, some authors (Mayer, 1988; markets.
Stiglitz, 1993) have even shown that competi- While these different studies have confirmed
tion between banks can be harmful to financial the general idea of the existence of a positive
stability and that a developed stock market is relationship between financial development and
not really important for financing the economic economic growth, they did not however detect
growth.
the direction of this relationship and whether Moreover, the failure of financial liberaliza- the financial development leads or follows the
tion process in some countries (Rabemanan- real output growth. Patrick (1966) who has in-
jara, 1998, p.18), and the inconclusive results vestigated this problem, was the first to iden-
of some studies (Ben Naceur and Ghazouani, tify two-way causality depending on the level
2005, p.219; Andersen and Trap, 2003, p.190 of economic development. The first one, called
question the positive relationship between fi- “demand following”, is from "financial" sec-
nancial development and economic growth. tor to the "real" one and characterizes the early
Meanwhile, some works have demonstrated stages of development. The second causal re-
that the relationship between financial develop- lationship between financial development and
ment and economic growth can be non-linear economic growth is termed “supply leading”. It
(Greenwood and Jovanovich, 1990, p.1076; is from the "real" sector to the "financial" one
Berthélemy and Varoudakis, 1996, p.11; Dei- and marks the mature stage of economic devel-
dda and Fattouh, 2002, p.339. In particular, it opment.
was shown that this relationship can be circular Otherwise, the nature of indicators measur-
or characterized by the presence of threshold ing financial development influences signifi-
effects.
cantly the intensity and the sense of its relation- More recently, several empirical studies ship with economic growth. The ambivalent
have focused on "too much finance" issue in findings of Gupta (1984) and Jung (1986) are
order to show the "Dark Side" of financial de- very eloquent at this level. While Gupta con-
velopment. In particular, they have investigated firms Patrick’s findings using M2/GDP as an
if there is a threshold above which financializa- indicator of financial development, the results
tion starts to have a negative impact on growth. obtained by Jung differ depending on the indi-
In this respect, Cecchetti and Kharroubi (2012) cator chosen. Indeed, the use of M1/GDP sup-
have examined how financial development af- ports this result whilethe use of M2/GDP does
fects aggregate productivity growth based on not.
a sample of 50 advanced and emerging econo- Similarly, because of the institutional and
mies during the period 1980-2009. They have political features of each country, the direction
found that, in the case of advanced countries, of causality differs from one system to anoth-
when private credit grows to the point where er, as observed by Demetriades and Hussein
it exceeds GDP, it becomes a drag on produc- (1996) and Demetriades and Aristis (1997).
tivity growth (Cecchetti and Kharroubi, 2012, Besides that, other studies have highlighted the
p.2). More generally, they have demonstrated fundamental difference between the short-term
that when the financial sector represents more dynamics and the long-term economic growth
than 3.5% of total employment, further increas- and financial development relationship (Fow-
es in financial sector size tend to be detrimental owe, 2010; Darart, 1999).
to growth. Arcand, Berkes and Panizza (2012), In contrast with the previous literature, some
focusing on the non-monotone relationship authors remain skeptical to the role of finan-
between financial and economic development cial systems in economic growth. Thus, Rob-
highlighted especially by Deidda and Fattouh
Chatri and Maaruf
(2002) and Rioja and Valev (2004), have found services and more specifically, financial inter- comparable results, using different data sets and
mediation function (Beck, 2011, p. 50). From empirical approaches. In particular, their results
this point of view, financial regulation and pol- suggest that finance starts having a negative ef-
icy should focus on the improvement of vari- fect on output growth when credit to the pri-
ous services provided by the financial system to vate sector reaches 80-100% of GDP (Arcand,
the real economy (payment, savings, credit and Berkes and Panizza, 2012, p. 6).
risk management) that may positively impact This non-monotonic relationship between
growth.
the size and growth of the financial sector is consistent with the vanishing effect highlighted
Research Method
by Rousseau and Wachtel (2011). Indeed, these authors have demonstrated that finance is not
Data
always good for growth. "Too much Finance" could conversely produce a “leakage effect”
The common problem faced by the majority of financial depth on the real economy. In this
of empirical works is to find appropriate indica- respect, they have demonstrated that credit to
tors to gauge the extent of financial develop- the private sector has no statistically significant
ment. In this paper, we opt for the traditional impact on GDP growth over the 1965-2004
indicators, largely inspired by the synthesis of period. Moreover, according to these authors,
Pagano (1993), which demonstrated that the fi- the excessive financial deepening or too rapid
nancial sector can affect economic growth via credit growth could weaken banking system
three channels: i) improving the Savings rate, and cause financial crisis and thereby could
ii) increasing savings allocated to productive have major negative implications for the real
investment and iii) reducing the cost financial economy.
resources as a result of the rise of the produc- In the same vein, De la Torre and al. (2011) 4 tivity of capital and reduction of risks . We will
focus also on the finance-crisis nexus. In this add, as well, an indicator reflecting the depth of respect, they have provided numerous insights
financial markets. Finally, insofar as the sources on the dangers of excessive financial develop-
of economic growth are not exclusively finan- ment. In particular, they have pointed out that
cial, the model discussed above also includes a the "Too much finance" result may be consistent
real variable control (investment). with positive but decreasing returns of financial
Our general model is written as follows: depth which, at some point, become smaller than the cost of instability brought about by the
GDPc t = α 0 + α 1 Z it + α 2 I t + ε t 6) dark side (De la Torre and al., 2011, p.2). These recent works presented above (among
with:
others) as well as the painful experience of the - GDPc t represents the Gross Domestic Prod- financial crisis of 2008 has led both academ-
uct per capita in constant prices as endog- ics and policymakers to reconsider their prior
enous variable, reflecting the development conclusions. It is nowadays largely admitted
of the real economy;
that financial development has also dark side. - Z it denotes a vector of financial variables: In particular a bloated financial system can be-
• The credit to private sector to GDP ratio come a drag on the rest of the economy.
(denoted CREDSP_GDP), measuring ef- It is for this reason that some authors believe
ficiency of the banking system in the allo- that the debate should focus not on the size of
cation of financial resources, as the most the financial sector and its impact on economic
of the funding of productive activities is growth, but rather on the benefits of financial 5 allowed by banks ;
4 This function of risk reduction is considered by some authors as rationale for financial intermediaries. Cf Allen, F.A. and Santomero,A, (1996), "The Theory of Financial Intermediation" Working paper, Wharton Financial Institutions Center.
5 The annual average bank credit to nominal GDP has increased from 49% in 2008 to 81% in 2011.
INDONESIAN CAPITAL MARKET REVIEW • VOL.VI • NO.1
• The liquidity ratio (denoted M3_GDP), + α 4 L (CAP - GDP) t calculated as the ratio of average aggre-
+ α 5 L (INVEST - GDP) t + ε t 7) gate broad money (M3) to GDP. This in-
dicator provides information on the size Remember that our aim in this paper is to test and the depth of the financial sector and
empirically the long-term relationship and the reflects also the banks efficiency in the
causality relationship between financial devel- mobilizing financial savings;
opment and economic growth in the particular • The ratio of market capitalization, which
context of Morocco. For this purpose, we use is equal to the values of listed securities
the Vector Error Correction Model (VECM), stock exchange divided by GDP (denoted
which has the advantage of incorporating the CAP_GDP). It is assumed that since the
concept of cointegration and distinguishes be- financial market reform of 1993, the stock
tween the short-term and long-term dynamics. market could be an additional source of
This method involves the following three steps: financing the economy;
The unit root test in order to estimate the • The interest rates on Treasury bonds 5
order of integration of the series, knowing that years (denoted rBDT 5Yrs ), chosen as refer-
most economic and financial series are non- ence for capturing the evolution of inter-
stationary.
est rate. In fact, this maturity is consid- The cointegration test for checking the exist- ered as benchmark line on the Moroccan
ence of a long-term stable relationship between market and therefore able to reflect the
non-stationary variables integrated of the same evolution of financial market conditions;
order. There are several tests for cointegration, - I t denotes the ratio of investment to GDP
the most known is Johansen test (1988), which (denoted INVST_GDP). It is used as a con-
is based on the maximum likelihood method. trol variable. We assume that the volume of
The causality test as any cointegrated sys- investment is positively correlated with eco-
tem implies the existence of an error correction nomic growth.
mechanism that prevents variables to deviate - ε t is the error term.
too much from their long-run equilibrium. Em- Although each individual financial indicator
pirically, the causal relationship can be ana- has its limits. Nevertheless, using this table of in-
lyzed using the error correction model (ECM) dicators provides an insightful picture of finan-
or the vector error correction model (VECM). cial development. The variables are considered
The estimation of these models will be com- in logarithmic forms: L(GDPc), L(M3_GDP),
pleted by the Granger causality test (1988), L (CREDSP_GDP), L(CAP_GDP), L(rBDT 5Yrs )
which assumes that the knowledge of the his- and L(INVST_GDP). They are quarterly and
tory of one variable improves the estimate of cover the period from 1998 to 2011. It should
another variable.
be noted in this regard that given the absence of quarterly data on investment covering the en-
Result and Discussion
tire period, we used the interpolation procedure highlighted by Goldstein and Khan (1976) to
Unit Root Test
generate quarterly data missing.
A long-term relationship between several
Econometric Methodology
variables requires two conditions. First, the variables must be non-stationary and integrated
Given the foregoing, the model (6) above on the same order for avoiding spurious regres- can be written as follows:
sion problems. Secondly, their stochastic trends must be related, that is to say, there must be one L(GDPc)t = α 0 + α 1 L (M 3 - GDP ) t or more linear combinations of these non sta- + α 2 L (CREDSP - GDP) t tionary variables that are stationary.
+ α 3 L (rBDT 5Yrs ) t
Chatri and Maaruf
Table1. Unit root test
Order of integration L(GDPc)
-3.712132 I(1) 1 st difference
-10.11346* I(0) L(M 3_GDP)
-0.577684 I(1) 1 st difference
-10.95825* I(0) L(CREDSP_
1 Yes
Yes
0.151848 I(1) GDP)
-0.865875 I(1) 1 st difference
-6.460597* I(0) L(CAP_GDP)
-1.240580 I(1) 1 st difference
-7.553538* I(0) L(rBDT 5Year )
-1.848286 I(1) 1 st difference
-7.858560* I(0) L(INVEST
1 Yes
Yes
-1.782685 I(1) _GDP)
-0.842862 I(1) 1 st difference
-4.159870* I(0) Notes : Tests ADF: *** reject of the unit rootat 1%, ** reject of the unit root at 5%, * reject of the unit root at 10%.
1 Yes
Yes
Table 2. Akaike and Schwarz criteria (VAR models)
Therefore, first, we determine the order of information according to the criteria of Akaike integration of the different series through the
and Schwarz.
standard unit root tests. We use for this purpose As it turns out most often Akaike criterion the Augmented Dickey Fuller, with the null hy-
gives contradictory results to Schwarz ones. pothesis of the nonstationarity. The following
The suggested lags are respectively 5 and 1. table gives the results, which show that all se-
However, the results of the estimation models ries are non-stationary and first order integrat-
led us to adopt 5 lags on level. Therefore, the
ed, denoted I (1) at 1%. This result suggests a Johanson test was made on stationary series stable long-term relationship between the used
with a 4 lags. The results (Table 3) show that at variables.
5%, there are three long term relationships, at least in one direction between the explanatory
Johansen Cointegration Test
variables and the endogenous variable, which exhibit similar behavior over time.
The previous results show that there is a high probability of cointegration between se-
Error Vector Correction Model and Granger
lected variables and permit subsequently to ap-
Causality
ply the vector error correction representation. Thus, in order to ensure proper specification of
The error vector correction model includes these models, it is necessary to determine the
two relationships that link the financial sector number of lags (p) optimizing the provision of
and economic growth. The long term station-
INDONESIAN CAPITAL MARKET REVIEW • VOL.VI • NO.1
Table 3. Johansen test
Hypothesized
0.05 No. of CE(s)
Trace
0.05 Max-Eigen
Critical Value Prob.** None *
Eigenvalue
Statistic
Critical Value
Prob.**
Statistic
40.07757 0.0003 At most 1
33.87687 0.0018 At most 2
27.58434 0.0096 At most 3
21.13162 0.1934 At most 4
Table 4. Long-term relationship between economic growth and financial development
Variable
Coefficient
Std. Error
t-Statistic
Prob.
0.000000 M3_ GDP
0.000000 CREDSP_GDP
0.000600 CAP_GDP
0.000100 rBDT 5YrS
0.015500 INVEST_GDP
8.173374 Adjusted R-squared
0.984743
Mean dependent var
0.133541 S.E. of regression
0.983218
S.D. dependent var
-5.175296 Sum squared resid
0.017300
Akaike info criterion
-4.958294 Log likelihood
0.014964
Schwarz criterion
645.455700 Durbin-Watson stat
Prob(F-statistic)
0.000000
ary relationship and the short-term relationship. tests on residuals show that it is sufficiently ro- The later takes into account the self-correcting
bust, as evidenced by their independance, nor- mechanisms leading to the convergence to the
mality, stationarity and homoscedasticity (see steady state.
Table A1 in Appendix).
Therefore, the results obtained using the Long-term Equilibrium Relationship
OLS estimation are robust and confirm the long term equilibrium relationship between financial
The simplest way to estimate the long-term variables and real variables, especially since relationship between economic growth and fi-
all selected explanatory variables seem explain nancial indicators is to apply the ordinary least
significantly economic growth. More specifi- squares (OLS) regression to the equation (7)
cally, the signs of variables "capturing" the fi- above.
nancial development are those expected. This Before discussing the results given in Table
corroborates the theoretical assumption under
4 above, it is necessary to ensure its statistical which the financial sector plays an active role and therefore operational quality.
in economic growth, through its ability to mo- The adjustment quality of the model, as
bilize saving and it allocation more efficiently, measured by the coefficient of determination
and to reduce the cost of resources and diver- (adjusted R²), is significant. This high level may
sify funding for private operators. nevertheless be related to the presence of col-
Indeed, the liquidity ratio, measured by the linearity between the explanatory variables in
ratio of broad money to GDP, affects signifi- the model, especially since it includes different
cantly and positively economic growth, with an financial variables, which may be highly cor-
elasticity of 0.5. This means that a long term related. However, the high significance of all
economic growth is expected to increase by regression coefficients at 1% lead us to believe
5% when the liquidity ratio increases by 10%. that collinearity is not strong and, therefore, to
Similarly, the increase in funding provided to not “drop”some variables, particularly since
private sector has a positive and significant our goal is to "capture" simultaneously the three
influence on economic growth. The elasticity channels transmission from finance to growth
between these two variables is 0.12 and means highlighted by Pagano (1993). Furthermore, the
that the long term economic growth is expected
to increase by nearly 1.2% when the credit dis- tributed to private sector increases by 10%. In addition, the development of direct funding, an- other important indicator of financial develop- ment seems also affect significantly and posi- tively economic growth. However, in this case, the elasticity is very low not exceeding 0.04. More specifically, the improvement of market capitalization of 10% would raise the long term economic growth by 0.4%. Finally, with regard to the link between interest rates and economic growth, it seems that it is significant and nega- tive and in accordance to what is expected. The results show that an increase of interest rate of 10% would reduce the long term economic growth by around 0.7%. This result confirms the active role in the development of financial sector in economic growth, so it reduces the cost of resources, and thereby improves the long-term growth potential.
All these results, consistent with the theoret- ical literature, mean that the financial system, in particular financial intermediation, influences significantly the Moroccan economic growth. This positive role attests that the improvement of the productive and allocative efficiencies of the financial sector is verified in the Moroccan context.
In contrast, the results show that the increase of investment by 10% would affect negatively the long-term economic growth (2.4%).This un- expected finding may nevertheless be explained by the following complementary features. First, on a general level, several studies on the effec- tiveness of investment in Morocco show that is non-productive (World Bank, 2006; Sekkat, 2004; Abouch and Ezzahid, 2004; Ministry of economy and Finance, 2002). More recently, the IMF in its Article IV consultations report of 2011, underlines clearly the persistence of low efficiency of investment projects (IMF, 2011, p.16). Furthermore, the analysis of the nation- al investment behavior on the period studied shows that it was supported in particular by public investment, along with the policy of the major projects undertaken by Morocco to catch
the recorded deficit in basic infrastructure 6 . The potential benefits of this kind of investment
cannot be felt only in the long term (Agénor, 2006). However, the model chosen covers a rel- atively short period (1998-2011) and thus is not able to capture entirely those effects. Moreover, the effort provided in term of investment in in- frastructure has benefited mainly to the rest of the world and not to the domestic productive sectors. Indeed, the imports to GDP ratio have increased from 28% in 1998 to nearly 49% in 2011.
After analyzing the results of the Johansen cointegration test and highlighted the long-term relationship between the growth and the finan- cial variables, we turn to estimate the error cor- rection model. This model permits to analyze, on the one hand, the speed of convergence of growth rate to its long term equilibrium level and, on the other hand, the contribution of fi- nancial sector in the short-term dynamics of economic growth.
Short-term Relationship
Given the optimal lag of our cointegration relationship, the basic model to estimate takes the form below. It should be noted that we can in the same way specify and estimate models in which the endogenous variable is represented by one of financial variables of our model.
ΔL(GDPc) t =
α i (ECMi) t-1
+ β i ΔL(GDP)i t-i
+ γ i ΔL(M 3 - GDP )i t-i
+ γ i ΔL(CREDSP - GDP)i t-i
+ ρ i ΔL(CAP - GDP)i t-i
+ τ i ΔL(rBDT 5Yrs )i t-i
+ φ i ΔL(INVEST - GDP)i t-i
+ ω+ε t
Chatri and Maaruf
6 Public investment has multiplied by more than seven from 1998 to 2011.
INDONESIAN CAPITAL MARKET REVIEW • VOL.VI • NO.1
Table 5. Error correction model-Wald test
D_L(CREDSP_GDP) D_L(CAP_GDP) D_L(rBDT 5Yrs ) Test Stat
D_L(GDPc)
D_L(M3_GDP)
Value Value
(P-value) (P-value) D_L(GDPc)
10.324 (0.0353) 1.8503 (0.7633) D_L(M3_GDP)
3.2718 (0.5134) 7.9837 (0.0922) D_L(CREDSP_GDP) F-statistic
9.2407 (0.0554) 2.0660 (0.7236) D_L(CAP_GDP)
3.0083 (0.5564) 4.5334 (0.3386) D_L(rBDT 5YrS )
13.416 (0.0094) 1.6887 (0.7928) D_L(INVEST_GDP) F-statistic
with: usually seen in underdeveloped countries, the ECM t-1 : represents the error correction term of
value of the speed adjustment appears rela- each cointegrating relationship lagged one time
tively low suggesting that after a shock, growth period. It is estimated from long term equilib-
tends to return rapidly towards its stationary rium. This term is denoted by Eviews by "Coin-
long term equilibrium. From a financial stabil- tEq."
ity point of view, this result reflects the fact that ε t : Random variable identically and indepen-
a temporary financial sector dysfunction is not dently distributed.
very troublesome in term of economic perfor- The results provided in Appendix (Table
mance because the return of the national econ- A2), which presents five different models, ac-
omy at its potential level should be quick. This cording to the variable taken as endogenous
result may be related to the fact that the stable variable. The error correction term has a fun-
relationship cointegration contains only market damental importance. It measures the speed of
capitalization and interest rates that are known adjustment of the dependent variable to its long
for their volatility and may therefore quickly re- run equilibrium level. More specifically, the co-
turn to their fundamental behavior. efficient should be negative, but less than -1 for
Otherwise, the long-term financial deep- long-term stable relationship.
ening seems to be "caused" by real variables, The results show that for each model there
except where it is measured by the volume of is at least one negative and statistically signifi-
loans to the private sector. Indeed, the results cant error term, which validates the existence
show that in this case the error term is not be- of a long-term relationship between financial
tween -1 and 0 (model 3). In contrast, at least deepening and economic growth. More pre-
one error term is negative and statistically sig- cisely, this result indicates that there is a restor-
nificant for the other measures. Thus, nearly ing force and error correction mechanism that
76% deviation of the liquidity of the economy brings financial and real variables to the same
compared to its long-run equilibrium is correct- long-term expansion path. It also means that se-
ed by the system (model 2). This rate drops to lected financial variables cause long-term eco-
23% if the cointegrating vector contains loans nomic growth.
(model 3). Similarly, the model corrects 73% of The estimation of the first model, which re-
the deviation of the market capitalization rela- tains the GDP per capita as the dependent vari-
tive to its long-term equilibrium level (model able revealed that this adjustment coefficient is
4). Finally, with regard to the interest rate, the -0.63 which means that the system corrects at
system corrects 67% of its deviation from its long term 63% of GDP per capita deviation. In
long-term equilibrium level (model 5). other words, after any shock, the rate growth
In addition, the Wald test (Table 5) shows returns to its long term equilibrium level in less
that short-term causality between economic than 6 months on average. Contrary to what is
development and financial deepening is true
Chatri and Maaruf
Table 6. Information criteria Akaike and Schwarz: VAR models for different lags
Table 7. Granger Test
F-Statistic Probability D_L(GDPc) does not Granger Cause D_L(M3_GDP)
Null Hypothesis:
Obs
54 3.21607 0.07885 D_L(M3_GDP) does not Granger Cause D_L(GDPc)
1.34304 0.25190 D_L(GDPc) does not Granger Cause D_L(CREDSP_ GDP)
54 3.40074 0.07098 D_L(CREDSP_ GDP) does not Granger Cause D_L(GDPc)
1.93094 0.17069 D_L(GDPc) does not Granger Cause D_L(CAP_ GDP)
54 2.48943 0.12080 D_L(CAP_ GDP) does not Granger Cause D_L(GDPc)
4.19721 0.04565 D_L(GDPc) does not Granger Cause D_L(rBDT 5YrS )
54 0.68266 0.41252 D_L(rBDT 5YrS ) does not Granger Cause D_L(GDPc)
0.17532 0.67718 D_L(GDPc) does not Granger Cause D_L(INVEST_ GDP)
54 5.15132 0.02749 D_L(INVEST_ GDP) does not Granger Cause D_L(GDPc)
0.02170 0.88347 D_L(M3_ GDP) does not Granger Cause D_L(CREDSP_ GDP)
54 10.6958 0.00193 D_L(CREDSP_ GDP) does not Granger Cause D_L(M3_ GDP)
0.28244 0.59741 D_L(M3_ GDP) does not Granger Cause D_L(CAP_ GDP)
54 0.57291 0.45259 D_L(CAP_ GDP) does not Granger Cause D_L(M3_ GDP)
0.32205 0.57287 D_L(M3_ GDP) does not Granger Cause D_L(rBDT 5YrS )
54 0.21287 0.64649 D_L(rBDT 5YrS ) does not Granger Cause D_L(M3 _ GDP)
3.73205 0.05894 D_L(M3_ GDP) does not Granger Cause D_L(INVEST_ GDP)
54 5.80552 0.01962 D_L(INVEST _ GDP) does not Granger Cause D_L(M3_ GDP)
2.02325 0.16100 D_L(CREDSP_ GDP) does not Granger Cause D_L(CAP_ GDP)
54 0.16339 0.68775 D_L(CAP_ GDP) does not Granger Cause D_L(CREDSP_ GDP)
0.41185 0.52391 D_L(CREDSP_ GDP) does not Granger Cause D_L(rBDT 5YrS )
54 0.60292 0.44105 D_L(rBDT 5YrS ) does not Granger Cause D_L(CREDSP_ GDP)
1.02792 0.31543 D_L(CREDSP_ GDP) does not Granger Cause D_L(INVEST_ GDP)
54 3.52533 0.06616 D_L(INVEST_ GDP) does not Granger Cause D_L(CREDSP_ GDP)
0.40292 0.52842 D_L(CAP_ GDP) does not Granger Cause D_L(rBDT 5YrS )
54 5.6E-07 0.99940 D_L(rBDT 5YrS ) does not Granger Cause D_L(CAP_ GDP)
2.79090 0.10093 D_L(INVEST _ GDP) does not Granger Cause D_L(rBDT 5YrS )
54 0.21201 0.64715 D_L(rBDT 5YrS ) does not Granger Cause D_L(INVEST_ GDP)
1.13261 0.29223 D_L(INVEST_ GDP) does not Granger Cause D_L(CAP_ GDP)
54 0.00461 0.94616 D_L(CAP_ GDP) does not Granger Cause D_L(INVEST_ GDP)
only when the latter is measured by the market and Ranciere (2002, p.15), who present it as an capitalization. Moreover, the causality seems
explanation for the apparent contradiction be- to be bidirectional in this case. In fact, market
tween the literature on crises and the literature capitalization "cause" economic growth, while
on endogenous growth.
all other variables seem to have no short-term From the econometric point of view, the re- effect on it. Conversely, the economic growth
sults of different specifications adopted are here does not "cause" any short-term financial vari-
also significant (Table A3 in Appendix). Tests able, except market capitalization. This result is
on the residuals validate their independence, not surprising that other variables (In particular,
normality and homoscedasticity hypothesis. liquidity of the economy, loans to private sec-
Otherwise, the R² is at a reasonable level (rang- tor) may exert their effects on growth on long
ing from 58% to 76%) with respect to the spec- term perspective.
ificity of the model. Indeed, in the short-term We note simply that this difference in the
perspective, growth rate is a function of several nature of causality between the two sectors
variables that cannot all be identified properly according to the time horizon (short and long
by an econometric representation .
term) has already been implemented by Loayza
INDONESIAN CAPITAL MARKET REVIEW • VOL.VI • NO.1
Granger Causality Test that the financial system reacts a passively to the needs expressed by the real sector. In oth-
To better understand the link between fi- er words, there should be a minimum level of nancial development and economic growth, it
growth of the GDP per capita before observing is important to complete this work by a test of
really its benefits on the financial activity. By causality in the Granger sense between the fi-
crossing this threshold, it leads to modernize fi- nancial variables and real variables. To do this,
nancial sector, in particular through creation of we have first estimated different VAR models
an integrated financial market, which will not in order to determine the optimal number of
fail to exert its effects on economic growth. lags (Table 6).
This result is in line with the findings of The results of the Granger test (Table 7),
several theoretical works including Robinson made on the basis of the series in first differ-
(1952), who considers that the financial sec- ence, show the absence of any feedback loop,
tor simply follows economic dynamic. It is ie bidirectional causality between the variables
also comparable to the findings of some em- of our model.
pirical works (among others Rioja and Valev, Otherwise, the unidirectional sense of cau-
2004, p.127; Deidda and Fattough, 2002, p.339; sality between economic performance and fi-
Berthelemy and Varoudakis, 1996, p.11) which nancial development depends on the indicator
show that it takes to reach a certain level of in- used to identify the financial deepening. Thus, if
come per capita, so that it is possible to develop financial development is measured by tradition-
the various types of intermediation and to ben- al indicators of intermediation – liquidity ratio
efit from their positive effect on the growth. or funding to private sector- the Granger cau-