proposal
mmemo
EMfinal.pdf
See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/271701336
Financial development and economic growth in an oil-rich economy: The case
of Saudi Arabia
Article in Economic Modelling · December 2014
CITATIONS
32 READS
236
3 authors:
Some of the authors of this publication are also working on these related projects:
ANP model for the NTP: The case of Saudi Arabia NTP2020 View project
Political economy of public goods provision View project
Nahla Samargandi
King Abdulaziz University
19 PUBLICATIONS 166 CITATIONS
SEE PROFILE
Jan Fidrmuc
Brunel University London
112 PUBLICATIONS 1,545 CITATIONS
SEE PROFILE
Sugata Ghosh
Brunel University London
44 PUBLICATIONS 387 CITATIONS
SEE PROFILE
All content following this page was uploaded by Jan Fidrmuc on 09 July 2018.
The user has requested enhancement of the downloaded file.
Financial development and economic growth in an oil-rich economy: The case of Saudi Arabia*
Nahla Samargandi†, Jan Fidrmuc‡ and Sugata Ghosh§
August 2014
Abstract
We investigate the effect of financial development on economic growth in the context of Saudi
Arabia, an oil-rich economy. In doing so, we distinguish between the effects of financial
development on the oil and non-oil sectors of the economy. Using the Autoregressive
Distributed Lag (ARDL) bounds test technique, we find that financial development has a
positive impact on the growth of the non-oil sector. In contrast, its impact on the oil-sector
growth and total GDP growth are either negative or insignificant. This suggests that the
relationship between financial development and growth may be fundamentally different in
resource-dominated economies.
Keywords: Financial Development; Economic Growth; ARDL Method; Oil and Non-oil
Sectors; Saudi Arabia.
JEL Codes: O11, O16, O47
* We benefited from helpful comments received from Mauro Costantini and workshop participants at Brunel University and Ruhr University-Bochum. We are also grateful to an anonymous referee for constructive comments and suggestions, which have improved the paper. Nahla Samargandi’s research on this paper was supported by a PhD scholarship from King Abdul Aziz University, whose financial support she would like to gratefully acknowledge. † Department of Economics and Finance, Brunel University; and Department of Economics, Faculty of Economics and Administration, King Abdulaziz University, Saudi Arabia; E-mail: [email protected]. ‡ Corresponding Author: Department of Economics and Finance and Centre for Economic Development and Institutions (CEDI), Brunel University; Institute of Economic Studies, Charles University; and CESifo Munich. Contact information: Department of Economics and Finance, Brunel University, Uxbridge, UB8 3PH, United Kingdom. Email: [email protected] or [email protected]. Phone: +44-1895-266-528, Web: http://www.fidrmuc.net/. § Department of Economics and Finance, CEDI and BMRC, Brunel University, UK; Email: [email protected]; Tel: +44 (0)1895 266887
2
I. Introduction
In this paper, we explore the link between financial development and economic growth in an
oil-rich economy, Saudi Arabia. To the best of our knowledge, our paper is one of the first
studies to specifically consider the role that financial development plays in a resource-
dependent economy, and the potentially different effects that it may have on the resource-
extraction and conventional sectors of such an economy. Countries whose economies are
dominated by oil or other natural resources possess specific features not shared either by
industrialized or developing economies. A large fraction, often a lion’s share, of economic
activity is represented by resource extraction, characterized by low added value and often by
a high degree of state regulation. Economic performance is predominantly driven by the
prices of natural resources that are determined in world markers rather than by domestic
economic developments.
The literature on the relationship between financial development and economic growth is
voluminous. There is, however, no consensus view yet on either the nature of this
relationship or the direction of causality. Four different hypotheses have been proposed.
The first view is that financial development is supply–leading, in the sense that it fosters
economic growth by acting as a productive input. This view has been supported theoretically
and empirically by a large number of studies. One of the earliest contributions is by
Schumpeter (1911) who argues that the services provided by financial intermediaries
encourage technical innovation and economic growth. McKinnon (1973) and Shaw (1973)
were the first to highlight the importance of having a banking system free from financial
restrictions such as interest rate ceilings, high reserve requirements and directed credit
programs. Such policies tend to be prevalent in all countries, but are especially common in
developing ones. According to their argument, financial repression disrupts both savings and
investment. In contrast, the liberalization of the financial system allows financial deepening
and increases the competition in the financial sector, which in turn promotes economic
growth. Similar ideas are put forward by, among others, Galbis (1977), Fry (1978),
Goldsmith (1969), Greenwood and Jovanovic (1990), Thakor (1996), and Hicks (1969). They
view financial development as a vital determinant of economic growth, which increases
3
savings and facilitates capital accumulation and thereby leads to greater investment and
growth.
Empirically, several studies support the supply–leading view. A prominent contribution is
King and Levine (1993). They study 80 countries by means of a simple cross-country OLS
regression. Their findings imply that financial development is indeed an important
determinant of economic growth. Similar results have been found by Chistopoulos and
Tsionas (2004), who examine the long-run relationship between bank development and
economic growth for 10 developing countries. They utilize panel cointegration techniques
and find a uni-directional relationship going from financial development to economic growth.
Atje and Jovanovic (1993) assess the role of the stock market on economic growth and find
that the volume of transactions in the stock market has a fundamental effect on economic
growth. Subsequent studies confirm these results by focusing on both market-based and
bank-based measures of financial development (see for example, Levine and Zervos, 1998,
and Kunt and Maksimovic, 1998).
The second view is demand-following. In contrast to the previous position, Robinson (1952)
argues that financial development follows economic growth, which implies that as an
economy develops the demand for financial services increases and as a result more financial
institutions, financial instruments and services appear in the market. A similar view is
expressed by Kuznets (1955), who suggests that as the real side of the economy expands and
approaches the intermediate stage of growth, the demand for financial services begins to
increase. Hence, financial development depends on the level of economic development rather
than the other way around. This view has also been empirically confirmed by several studies
such as Al-Yousif (2002) and Ang and McKibbin (2007).
The third view is one of bidirectional causality. Accordingly, there is a mutual or two-way
causal relationship between financial development and economic growth. This argument was
first put forward by Patrick (1966) who posits that the development of the financial sector
(financial deepening) is as an outcome of economic growth, which in turn feeds back as a
factor of growth. Similarly, a number of endogenous growth models such as Greenwood and
Jovanovic (1990); Greenwood and Bruce (1997); and Berthelemy and Varoudakis (1997)
4
posit a two-way relationship between financial development and economic growth.
Additional support for this view can be found in the empirical study by Demetriades and
Hussein (1996), who studied 13 countries and found very strong evidence supporting
bidirectional causality.
Finally, the fourth view states that financial development and economic growth are not
causally related. Based on this view, financial development does not cause growth or vice
versa. This view was initially put forward by Lucas (1988) who states that “economists badly
overstress the role of financial factors in economic growth”. His view is also supported by
Stern (1989).
In addition, some empirical studies of the effects of financial development on economic
growth highlight the potential negative association between finance and growth. For example,
De Gregorio and Guidotti (1995) find a negative impact of financial development on growth
in some Latin American countries. Van Wijnbergen (1983) and Buffie (1984) also point out
the potentially negative impact of finance on growth. They argue that the high level of
liberalization of the financial sector (financial deepening) results in decreasing the total real
credit to domestic firms, and thereby lowers investment and slows economic growth. Al-
Malikawi et al (2012), who examine the short- and long-run relationship between financial
development and economic growth in the United Arab Emirates (UAE), suggest the
relationship between them is negative. They attribute this result to the transition phase of the
UAE financial system during the period of study, as well as to the weak regulatory
environment of the financial intermediaries.
To the best of our knowledge, only few studies attempt to investigate the relationship
between financial development and economic growth in the context of a natural-resource
dominated economy.1 Nili and Rastad (2007), and Beck (2011), are among the few authors
who consider how the abundance of oil can affect the relationship between financial
development and economic growth, and whether there is any indication of a natural resource 1 A number of studies provide evidence that countries endowed with natural resources have a tendency to grow more slowly than less resource-abundant countries. This phenomenon is known as resource curse thesis (see Sachs and Warner, 2001; Nankani, 1979). Resource curse refers to the negative externalities stemming from the abundance of natural resources to the rest of the economy. See van der Ploeg (2011) for a recent survey on the curse of natural resource abundance.
5
curse in the relationship between financial development and economic growth. Nili and
Rastad (2007) examine the role that financial development plays in oil-rich economies. They
find that financial development has a weaker effect in oil-exporting countries than in oil-
importing countries. They suggest that this result is not only due to the high dependence on
oil in the former but also because of the general inefficiency of financial institutions in oil-
dependent countries. Beck (2011), in turn, argues that the ambiguity in the relationship
between financial development and economic growth in oil-rich (or natural-resource-rich)
countries in the previous literature reflects the fact that economic growth is driven by
different forces in these countries, and that the financial sector has a different structure and
plays a different role there. Nevertheless, his findings indicate, contrary to Nili and Rastad
(2007), that there is in fact no significant difference in the impact of financial development on
economic growth between resource-based countries and non-resource based countries.
However, when he assesses the level of countries’ reliance on natural resources, he finds that
countries that depend more on exports of natural resources tend to have underdeveloped
financial systems. This is despite the fact that banks in resource-based economies tend to
display higher profitability and are more liquid and better capitalized. However, they offer
less credit to the private sector, which he attributes to the incidence of financial repression in
resource-based countries. Therefore, he concludes that resource-based countries can be
subject to the natural resource curse in financial development.
We seek to contribute to this debate by considering the case of a resource-dominated country:
Saudi Arabia.2 The economy of Saudi Arabia is heavily dependent on oil revenue. Recently,
however, the government has been promoting diversification towards the non-oil sector and
reducing the country’s dependence on the petroleum sector. Since the implementation of the
fourth development plan (1985-1990), in particular, significant priority has been given to the
financial sector. We investigate, therefore, the role that the financial sector plays in this
2 Substantial literature focuses on single country studies, e.g., Murinde and Eng (1994) for Singapore; Abu- Bader,et al (2008) for Egypt; Lyons and Murinde (1994) for Ghana; Odedokun (1989) for Nigeria; Agung and Ford (1998) for Indonesia; Wood (1993) for Barbados; Khan, et al (2005) for Pakistan; Hondroyiannis , et al. (2005) for Greece; Ang, et al. (2007) for Malaysia; Majid (2007) for Thailand; Mohamad (2008) for Sudan; Singh (2008) for India; Safdari et al. (2011) for Iran; Thangavelu, et al. (2004) for Australia; Muhsin and Pentecost (2000) for Turkey; Qi Liang,et al (2006) for China; Ghatak (1997) for Sri Lanka and Al-Malikawi et al. (2012) for UAE.
6
country’s economy, and whether this role differs between the traditional sector (petroleum)
and the emerging non-oil sector.
To this effect, we collect time-series data from 1968 to 2010 and apply an ARDL bound test
approach to cointegration to examine the long and short-run impact of the financial sector on
economic growth. There are various methods for examining the existence of a long-run
relationship between the variables of interest: Engle and Granger (1987) and Johansen (1988,
1991, 1995) are the most widely adopted approaches. We, however, follow the ARDL bound
test approach for testing the finance and growth nexus due to the favorable features of this
technique compared to the other conventional approaches, as discussed in more detail in the
methodology section. Furthermore, we deviate from the usual approach by using principal
component analysis (PCA) to build a single composite indicator of financial development.
Our findings indicate that financial development has a statistically significant and positive
effect on the non-oil sector only. In contrast, the effect on overall GDP is either not statically
significant or negative and significant. We consider this an important result, not only from the
perspective of an oil-rich economy, but also in the general context of the financial
development-growth debate.
The remainder of this paper is organized as follows. Section II provides a brief overview of
the Saudi economy and discusses the key characteristics of its financial sectors. Section III
describes the data and the construction of the measures of financial development used in the
empirical analysis. Section IV explains the methodology and the econometric model used in
our study. Section V reports the empirical results. Finally, section VI concludes, and provides
some policy implications.
II. Overview of the Saudi Economy and its Financial Sectors
Saudi Arabia’s economy depends heavily on the oil sector. The country is the world’s leading
exporter of petroleum and a very prominent member of the OPEC. The oil sector accounts for
about 45 percent of the total GDP and 90 percent of the total export earnings. In order to
reduce the dependence on the oil sector, the government has, over the last couple of decades,
7
been trying to diversify the economy by promoting the non-oil sector. Efforts have been
made to diversify into power generation, telecommunications, natural gas exploration,
and petrochemical sectors. What is more, in order to foster economic growth, the
government has recognized the important role of the financial sector in mobilising savings
and channeling funds to economic activities. To this effect, it has been promoting the
development of an efficient banking system, well-developed financial markets and
comprehensive and competitive insurance services.
There have been several signs that the economy has been switching from the oil to the non-oil
sector over the last four decades.3 During the 1970s, the share of the non-oil sector in overall
GDP was very low, from 30% to 37%. However, at the beginning of the 1980s, the Saudi
economy experienced a rapid shift in favour of the non-oil sector at the expense of the oil
sector. In 1985, the non-oil output peaked at 77% of GDP. Thereafter, its share fluctuated
between 60% and 72% during the following period (1986-2010).
Choudhury and Al-Sahlawi (2000) see this significant growth of the non-oil sector as a
success of the emphasis on diversification made in the fourth development plan (1985-90)
and all the subsequent plans. On the other hand, Al-Hassan et al. (2010) argue that these
increases in the non-oil sector are merely the result of the fluctuation in the world’s oil
demand that reflects swings in world oil prices.
Although the financial sector in Saudi Arabia comprises both banks and non-bank financial
institutions, it is dominated by the banking sector. This is because all other financial
intermediaries and non-bank financial institutions, such as the stock market, Sukuk (Islamic
bonds) and insurance companies, are either newly-established or underdeveloped. For
example, the Saudi stock market was officially established only in 1984; until then it was just
an informal market. Moreover, the number of listed companies was small: just 72 companies
up to 2008.4
3 The oil sector refers to the production activity relating to the extraction and supply of crude oil. The non-oil activities include finance, trade, government services, construction, utilities, natural gas and petroleum- processing industries. 4 However, the Saudi stock market has experienced tremendous development in the last five years due to the new rules allowing non-Saudi citizens to participate in shares trading in the stock market which used to be
8
Although the Saudi insurance industry is the largest insurance market among the Gulf
Cooperation Council (GCC) countries, the regulation of this sector by the Saudi Arabian
Monetary Agency (SAMA) only began in 2003 (The Saudi Insurance Market Report, 2009).
In 2004, there was only one insurance company, but by the first half of 2008, the Council of
Ministers approved the licensing of 22 insurance companies. As regards the Islamic Banking
and Sukuk (Islamic bonds) sector, there are four Islamic banks in Saudi Arabia; in addition to
them, there are Islamic windows in the conventional banks. According to a report issued by
the World Islamic Banking Conference on the competitiveness of Islamic banks, Saudi
Arabia ranks first, as measured by the earnings of Islamic Banks over the period 2000–2006.
However, no data on this sector are publicly available.
The banking sector has fared well during the last four decades, no doubt favourably affected
by the oil boom phase. Several Saudi commercial banks were established and the number of
commercial banks has risen to 12. Out of those, five are entirely owned by Saudi
shareholders while the rest are owned by a mix of Saudi and foreign shareholders (Ariss, et
al., 2007). Table 1 shows some selected indicators of the banking sector. The ratio of liquid
liabilities to GDP (M3/GDP) has increased moderately from 2005 to 2010, though it has
fallen somewhat in 2008 and 2010 compared to the previous years. A higher liquidity ratio
means that the banking system has grown in size. The ratio of the private sector credit to
GDP has followed the same trend as the liquid liabilities to GDP ratio. Table 1 also shows
that total bank assets have been increasing constantly over the years.
The Saudi commercial banks have expanded the amount of investment and consumer
lending. The private sector in Saudi Arabia remains relatively small, possibly because it is
constrained by the limited credit disbursement by the commercial banks to the private sector.
However, more commercial banks entered into the money market and expanded their loans to
the private sector from 1999 onwards so that the loan disbursements have increased sharply.
Table 2 also shows that the total credit disbursement of commercial banks has increased
moderately from 2006 to 2010, but has fallen slightly in 2009 as compared to the previous
year. restricted only to Saudi citizens before 2008. As a result, more companies were encouraged to seek finance from the stock market and the number of listed companies increased to 172 companies in 2013.
9
III. Data and the construction of financial development variables
Data description
We use annual data for Saudi Arabia covering the period from 1968 to 2010. The data was
collected from the World Development Indicators (WDI) dataset and the 47th annual report
of the Saudi Arabian Monetary Agency (SAMA). The variables of interest include real gross
domestic product per capita (GDP) as the dependent variable and potentially important
determinants of economic growth as explanatory variables. We initially collected data on
government expenditure (as a percentage of GDP), investment share in GDP, oil price,
inflation, openness to trade and various measures of financial development (discussed in
greater detail below).5 However, when including all variables in the regression, several turned
out to be insignificant. We, therefore, proceeded to omit the insignificant explanatory
variables, one by one, until we were left with a model that contained only significant
variables: the oil price (OILP), trade openness (TRD) and financial development (FD).6 The
fact that investment dropped out is particularly puzzling: it is typically a robust determinant
of economic growth in most studies. The fact that it fails to feature significantly as a
determinant of Saudi growth may be due to the overwhelming dominance of the oil sector in
this country. It may also reflect the fact that a large fraction of investment in Saudi Arabia is
related to oil exploration and thus may affect growth only with a substantial lag, likely to be
several years.
We, therefore, estimate a model that includes only a relatively narrow set of core variables
alongside our main variable of interest: financial development. This is in line with the
literature arguing against controlling for a relatively extensive list of explanatory variables:
the resulting coefficients then often depend crucially on the set of specific remaining
variables included (see the discussion in, among others, Levine and Renelt, 1992, and Woo,
2009).
5 We also sought to include some measure of human capital but were unable to do so because of a large number of missing values. 6 This approach is equivalent to implementing the general-to-specific procedure.
10
Construction of financial development variables: Principal component analysis (PCA)
We collected information on the following three indicators of financial development:
1. The ratio of broad money (M2)7 to nominal GDP.
2. The ratio of liquid liabilities (M3)8 to nominal GDP.
3. The ratio of credit to private sector to nominal GDP.
We follow Ang and McKibbin (2007) in constructing a single measure of financial
development by using principal component analysis. The justification for doing this is two-
fold. First, it addresses the problem of multicollinearity, or the high correlation between the
various financial development indicators. Second, there is no general consensus as to which
measure of financial development is most appropriate. Therefore, having a summary measure
of financial development that includes all the relevant financial proxies (data permitting) to
capture several aspects of the financial sector at the same time, such as directed credit
programs and liquidity, will provide better information on financial deepening.
Table 3 presents the results of principal component analysis with the logarithms of the three
measures of financial development listed above. The eigenvalue associated with the first
component is significantly larger than one. The first principal component explains
approximately 97.3% of the standardised variance, the second principal component explains
another 2.0%, and the last principal component accounts for only 0.5% of the variation.
Clearly, the first principal component is the best measure of financial development in this
case. Below, we denote this summary indicator of financial development as FD.
IV. Methodology and Model Specification
Methodology
The two commonly used techniques to test for cointegration between variables are the Engle
and Granger method and the Johansen technique. The Engle and Granger method is a single-
7 M2 = M1 (currency outside banks + demand deposits) + time and saving deposits. 8 M3= M2 + other quasi monetary deposits.
11
equation technique and as such it can lead to contradictory results, especially when there are
more than two cointegrated variables under consideration (see, Asteriou and Hall (2011);
Ang (2010)). Another shortcoming of this method is in its implementation: in order to obtain
the long-run equilibrium relationship, we need to estimate the Ordinary Least Squares (OLS)
regression as a first step. This procedure, as pointed out by Banerjee et al. (1986), may
generate a substantial bias owing to the omission of dynamics and this can undermine the
performance of the estimator. Also, the two-step residual-based procedure uses the generated
residual series in the first step to estimate a new regression model in the second stage, in
order to see whether the residual series is stationary or not. Hence, the error introduced in the
first step is carried forward into the second step (Enders, 2004; Asteriou and Hall, 2011).
The Johansen method, which is known as a system-based approach to cointegration, is
considered to be a superior method over the Engle and Granger method, and offers a solution
in the case of having more than two variables and multiple cointegration vectors that might
exist between the variables. Furthermore, the Johansen approach mitigates the omitted lagged
variable bias that affects the Engle and Granger approach by the inclusion of lags in the
estimation. Even so, the Johansen method can be subject to criticism. The first drawback is
the sensitiveness of the results to the optimal number of lags included in the test (Gonzalo,
1994). The second is that if there are more than one cointegrating vectors, it is often hard to
interpret each implied economic relationship and to find the most appropriate vector for the
subsequent test (Ang, 2010).
Both the Engle-Granger and Johansen techniques are criticised on the grounds that the
validity of these methods requires that all the variables be integrated of order one, I(1). They
cannot be employed, therefore, if we have a mixture of I(0) and I(1) variables, as in our case
(see below).
In this study, we use the autoregressive distributed lag or Bounds testing approach to
cointegration (ARDL) technique of Pesaran et al. (2001). This method has been used as an
alternative cointegration test that examines the long-run relationships and dynamic
interactions among the variables and as such addresses the above issues. This approach has
several desirable statistical features. First, the cointegrating relationship can be estimated
12
easily using OLS after selecting the lags order of the model. Second, it allows to test
simultaneously for the long and short-–run relationship between the variables in a time series
model. Third, in contrast to the Engle-Granger and Johansen methods, this test procedure is
valid irrespective of whether the variables are I(0) or I(1) or mutually co-integrated, which
means that no unit root test is required. However, this test procedure will not be applicable if
an I(2) series exists in the model. Fourth, in spite of the possible presence of endogeneity,
ARDL model provides unbiased coefficients of explanatory variables along with valid t-
statistics. In addition, ARDL model corrects the omitted lagged variable bias (Inder, 1993).
Furthermore, Jalil et al. (2008) and Ang (2010) argue that the ARDL framework includes
sufficient numbers of lags to capture the data generating process in general to specific
modelling approach of Hendry (1995). Finally, this test is very efficient and consistent in
small and finite sample sizes.
Model specification:
Following Ang and McKibbin (2005), Khan and Qayyum (2005) and Fosu and Magnus
(2006), the ARDL version of the vector error correction model (VECM) can be specified as: ∆ ln Yt=β0+β1 ln Yt-1+β2 ln X1 t-1+β3 ln X2t-1+β4 ln X3t-1 + ∑ γi∆ ln Yt-i +pi ∑ δj∆qj ln X1t-j+ ∑ φl∆ ln X2t-l + ∑ ηm ln X3t-m+εtqmql , (1) In equation (1), Y is the real gross domestic product per capita, X1 stands for financial
development, X2 is the oil price, X3 is trade openness, and ε is the error term.
Using the ARDL approach we estimate three models with the dependent variable being real
GDP per capita (GDP), real GDP per capita of Non Oil Sector (GDPN) and real GDP per
capita of Oil Sector (GDPO). Each of these is regressed on Financial Development (FD), Oil
Price (OILP), and Trade Openness (TRD).
Estimation procedure
We first estimate equation (1) using OLS and then conduct the Wald Test or F-test for joint
significance of the coefficients of lagged variables for the purpose of examining the existence
13
of a long-run relationship among the variables. We test the null hypothesis, (H0): = == = 0, that there is no cointegration among the variables, against the alternative hypothesis (Ha): ≠ ≠ ≠ ≠ 0. The F-statistics is then to be compared with the critical value (upper and lower bound) given by Pesaran et al. (2001). If the F-statistic is
above the upper critical value, the null hypothesis of no cointegration is rejected which
indicates that long-run relationship exists among the variables. Conversely, if the F-statistic is
less than the lower critical value the null hypothesis cannot be rejected, implying no
cointegration among the variables. However, if the F-statistic lies between lower and upper
critical values, the test is inconclusive.
In the second step, after testing the relationship among the variables, the long-run coefficients
of the ARDL model can be estimated:
ln Yt = + ∑ ln ∑ ln + ∑ ln + ∑ ln + , (2) In this process, we use the SIC criteria for selecting the appropriate lag length of the ARDL
model for all four variables under study. Finally, we use the error correction model to
estimate the short run dynamics:
∆ ln Yt = + ∑ ∆ ln ∑ ∆ ln + ∑ ∆ ln + ∑ ∆ln + + . (3) Cusum and cusumsq test (stability tests)
We perform two tests of stability of the long-run coefficients together with the short run
dynamics, following Pesaran (1997), after estimating the error correction model: the
cumulative sum of recursive residuals (CUSUM) and the cumulative sum of squares of
recursive residuals (CUSUMSQ) tests.
V. Results and Discussion
Unit-root test
Prior to testing for cointegration, we conduct a test of the order of integration for each
variable using the Augmented Dickey-Fuller test (Table 4). Even though the ARDL
framework does not require the pre-testing of variables, the unit root test could indicate
14
whether or not the ARDL model should be used. As can be seen from Table 4, only some of
the variables, in particular real GDP per capita in the non-oil sector (GDPN), real GDP per
capita in the oil sector (GDPO) and the oil price (OILP), are stationary at the 5 percent or 10
percent significance level, whereas all variables are stationary after first differencing. Hence,
the results of the unit root test demonstrate that the ARDL model is more appropriate to
analyze the data than the Johansen cointegration model.
Cointegration test
The calculated F-statistics for the cointegration test are displayed in Tables 5, 9 and 13. The
F-statistic for the first model (7.5803, Table 5) is higher than the upper bound critical value at
the 1 percent level of significance, using restricted intercept and no trend. This implies that
the null hypothesis of no cointegration cannot be accepted, therefore there is a cointegrating
relationship among the variables. Through normalization process we find that there is
cointegration at 5% when financial development and the oil price are the dependent variables
but not when we consider openness to trade. The same procedure has been applied to analyze
the other two models (for the oil and non-oil sectors). The results suggest the presence of
cointegration between GDPN and all explanatory variables, and also cointegration between
GDPO and the explanatory variables.
Long- run impact
The empirical results are reported in Tables 6, 10 and 14. They show that trade openness has
a positive and significant effect on overall economic growth as well as on the growth of both
oil and non-oil sectors. This result is consistent with theoretical and empirical predictions. In
addition, the oil price has a positive and significant impact on overall GDP growth but an
insignificant impact on the non-oil sector in the long-run.
Financial development has a negative but insignificant impact on economic growth,
indicating that the Saudi economy has not benefitted from financial development. This result
is in line with Barajas, Chami and Yousefi (2012), who find that financial development has
lower if not negative effect on economic growth in oil-rich and Middle Eastern and North
African (MENA) countries. This finding may be attributed to the fact that during the period
15
under analysis, the financial sector was still relatively under-developed, and below a certain
threshold, beyond which it would be capable of promoting economic growth (Al-Malkawi et
al., 2012). Ram (1999) also found a negligible or weak negative impact of financial
development on economic growth. Jalil and Ma (2008), similarly, argue that inefficient
allocation of resources by banks coupled with the absence of favourable investment
environment in the private sector slow the overall economic growth in China. The findings of
Jalil and Ma would be applicable to Saudi Arabia where, as in China, most economic
decisions are directed by the government. Barajas et al. (2011) argue that the impact of
financial deepening on economic growth disappears in the case of an oil-based economy like
Saudi Arabia. Our findings are in line also with Ang and McKibbin (2006) who find no
evidence of economic improvement due to expansion of the financial sector in Malaysia. Ang
and McKibbin suggest that the returns from financial development depend on the
mobilization of savings and allocation of funds to productive investment projects. However,
due to information gaps, high transaction costs and improper allocation of resources, the
interaction between savings and investment and its link with economic growth is not strong in
developing countries. According to Beck (2011), the existence of natural resource curse in
financial development might be another reason for this insignificant impact of financial
development on growth in oil-rich economies.
In contrast, the effect of financial development (FD) on the non oil sector in Saudi Arabia is
positive and statistically significant at 10%. The magnitude of this impact is not sufficient to
ensure a positive relationship for the overall economy since the non-oil sector constitutes
only a relatively small part of the Saudi economy. This finding is consistent with Nili and
Rastad, (2007) who find that financial markets in resource-rich countries are relatively weak.
They attribute their results to three reasons, a possible natural resource curse in financial
development, the dominant role of government in total investment and the poor performance
of the private sector in these countries.
In contrast, the third model shows that FD does not have any impact on the oil sector of Saudi
Arabia. Since the oil sector is exclusively controlled by the government, it is not surprising
that financial development does not significantly contribute towards its growth.
16
Short run impact and adjustment
The coefficients of the error correction model for all three specifications are presented in
Tables 7, 11 and 15. The negative signs of each coefficient of the ECM variable reveal that
short-run adjustment, which occurs at a high speed in the negative direction, is statistically
significant. Moreover, this is an indication of cointegration relationship among GDP (both oil
and non-oil), financial development, oil price, and trade openness. The values of ECM
coefficients strongly suggest that the disequilibrium caused by previous year’s shocks
dissipates and the economy converges back to the long-run equilibrium in the current year
(see Dara and Sovannroeun, 2008; and Hossein, 2007).
Diagnostic test
The overall goodness of fit of the estimated models shown in Tables 8, 12 and 16 is quite
high, with R2 values of 96%, 99% and 77% for the first, second and third model, respectively.
This is not surprising, given that the ARDL model includes the lagged dependent variable.
We applied a number of diagnostic tests to the ARDL model. We found no evidence of serial
correlation, multicollinerarity, and error in the functional form, but found heteroskedasticity
in model 2 and model 3 (Tables 12 and 16). However, as Shrestha and Chowdhury (2005)
and Fosu and Magnus (2006) point out, it is natural to detect heteroskedasticity in the ADRL
approach, since the model mixes time series data integrated of order I(0) and I(1). Figures 1,
2 and 3 show the CUSUM and the CUSUMSQ stability test results to the residuals of
equation (1): the CUSUM and CUSUMSQ remain within the critical boundaries for the 5%
significance level. These statistics confirm that the long-run coefficients and all short-run
coefficients in the error correction model are stable and affect growth.
Robustness checks
Although the three previous models have passed all diagnostic and stability tests successfully,
we also carry out a number of robustness checks in order to examine the sensitivity of our
17
findings to alternative model specifications. In this section, we report the core results of these
robustness checks.
First, we re-estimate all models with the individual measures of financial development
variables (M2, M3 and credit to the private sector, all as fractions of GDP) individually rather
than as a composite index. The results are similar to those reported above in that the effect of
the financial development variable on growth is either negative and significant or
insignificant. Most notable result with the separate measures of financial development is that
the impact of claims on the private sector to GDP always appears to have a negative and
significant effect on economic growth. This finding suggests that there are fundamental
problems of credit allocation in the Saudi financial sector, due to the inefficient financial
regulation and supervision in the banking sector in Saudi Arabia, along with the lack of an
appropriate investment climate required to foster private investment and promote economic
growth in the long-run. Using (M3/GDP) and (M2/GDP) each in separate models along with
claims on private sector and other controls, we obtained positive and significant coefficients
in the long-run only for the growth of non-oil GDP model. To save space, we are not
reporting these results but they are available upon request as a supplementary appendix.
As a second robustness exercise, we consider another (non-money stock) variable used in the
literature as measure of financial development: total banks assets to GDP ratio. This variable
is a comprehensive measure of the size of the financial sector relative to the size of the
economy as whole (Levine and Beck, 1999). The total banks assets include claims on the
government, public enterprises and the private sector. Since we use claims on the private
sector as another measure of financial development, we exclude this variable from the total
banks assets. We denote the resulting measure as TBA.
We use TBA to replace M2/GDP. As discussed before, monetary aggregates such as M2 and
M3 as ratios of nominal GDP are the two most commonly used measures to capture the depth
of the financial sector, as used in the empirical literature. The reason for dropping M2/GDP is
that it has been argued in the literature that M2/GDP might not be that good a proxy for
financial development in the case of developing countries (e.g., Demetriades and Hussein,
18
1996; and Luintel and Khan, 1999) because currency held outside the banking system is a
large component of the broad money stock (M2) in these countries. If this is the case, an
increase in the ratio of broad money to GDP may reflect more extensive use of currency
rather than an increase in the volume of bank deposits. As a result, M2 mostly represents the
ability of the financial systems to provide transaction services rather than their ability to link
up surplus and deficit agents in the economy. Therefore, we omit M2/GDP and replace it
with TBA/GDP.
We apply the same principal component analysis procedure as before to construct a new
aggregate index of financial development. We denote this new summary indicator as FD2.
Hence, we aggregate the following three different measures of financial development into a
single index:
1- The ratio of liquid liabilities (M3) to nominal GDP.
2- The ratio of credit to private sector to nominal GDP.
3- The ratio of the total banks assets to nominal GDP.
Table 17 presents the results obtained from principal component analysis of the three
measures of financial development listed above. The first component explains 96% of the
variance in the data and its eigenvalue is larger than one. The second and third principal
component each explain only a negligible share of the variation. As before, we therefore, use
only the first principal component as a measure of financial development.
Robustness checks using FD2 index. 9
Cointegration test
The F-statistics for the cointegration tests are presented in table 18, 22 and 26. The F-statistic
of the models estimated with GDP, GDPN and GDPO are 6.763, 7.4093 and 4.837,
respectively, greater than the upper bound Pesaran critical value (4.37) at the 1 percent
9 We also carry out separate analyses using each of the original financial development indicators. The results are similar to those in Tables 18 to 29. In order to conserve space, we drop them from this version and make them available upon request.
19
significance level for the overall GDP and the non-oil sector and at 5 percent significance
level for the oil sector, using restricted intercept and no trend. This suggests that there is a
long-run relationship among the total GDP and the two sub-components of the total GDP;
GDPN and GDPO with the financial development index and the other two controls variables:
oil price and trade. Thus, the results imply that there is a unique cointegrating relationship
among the three dependant variables; GDP, GDPN, GDPO, and the explanatory variables.
Long- run impact
The existence of a long run relationship among GDP (both oil and non-oil) and the
explanatory variables allows the estimation of long run coefficients and short run dynamic
parameters. The empirical results of the long-run impact are presented in Tables 19, 23 and
27. The results for the control variables, oil price and trade confirm our previous findings.
The new financial development index displays a negative impact on long-run overall growth
and the growth of the oil sector, but this is now statistically significant. This finding is in line
with Mahran (2012), who finds a negative impact of the banking sector on the overall GDP
growth. In contrast, financial intermediation positively affects the growth rate of the non-oil
sector.
Short-run impact and adjustment
The results of the short-run and the lagged error correction term (ECM) are reported in tables
20, 24 and 28. The coefficient of the ECM for GDP and GDPN models; -0.164 abd -0.366,
respectively, are negative and statistically significant at the 1 percent level. The coefficient
for GDPN is also negative but significant at 10% only. The significant negative signs of all
ECM coefficients are an indication of a cointegrating relationship among real GDP (both
GDPN and GDPO) and financial development, oil price and trade and any disequilibrium
caused by previous year’s shocks converges back to the long-run equilibrium in the current
year for all models.
20
Diagnostic tests
Tables 21, 25 and 29 display the diagnostic test results for the underlying ARDL equation.
The results suggest again that all models pass the diagnostic tests against serial correlation,
functional form misspecification and non-normal errors. However, the GDPN and GDPO
models fail the heteroscedasticity test at 5%. As discussed earlier, it is natural to detect
heteroscedasticity when we have mixed time series data integrated of order I(0) and I(1). The
plot of the cumulative sum of recursive residuals (CUSUM) and cumulative sum of squares
recursive residuals (CUSUMQ) for the three robustness models presented in Fig. 4, 5 and 6
also indicate stability in the coefficients over the sample period as they fall within the critical
bounds.
As discussed before, financial systems in Saudi Arabia can be broadly classified as bank-
dominated. However, following the preceding robustness checks, we investigate how our
benchmark results change when we consider not only bank sector effects but also stock
market effects in our models. We carried out these estimations on shorter time span (1985-
2010) as this is the period for which the data on the stock market are available. We add the
market value of shares/GDP as a stock market variable measuring the development in the
financial sector along with the other financial development variables used in the main
analysis. The results show that the inclusion of stock market development does not
remarkably change our results. This indicates that financial development has a positive short-
run impact on the growth of the non-oil sector in Saudi Arabia. However, this impact
disappears in the long-run. In contrast, the impact of financial development on total GDP
growth and oil-sector growth are negative but insignificant. The control variables have the
expected sign with more or less minor changes.10
In summary, we confirm that our previous results are robust to alternative model speciations.
Moreover, we can conclude that financial development has a positive impact on the growth of
the non-oil sector in Saudi Arabia. In contrast, its impact on the oil sector and overall GDP
growth is negative and significant.
10 The results on bank and market sectors are provided in a supplement that can be obtained from the authors upon request.
21
VI. Conclusions
This paper contributes to the literature on financial development and growth by focusing on
the financial sector of an oil-rich economy, Saudi Arabia, which has not been studied
extensively thus far. The results of this empirical study, based on the ARDL approach,
suggest that financial development has a positive impact on economic growth of the Saudi
non-oil sector in the long-run. In contrast, we find a negative or insignificant impact of
financial development on the economy as a whole, and on the oil sector, which we believe is
a significant finding.
These results can be interpreted from two angles. First, they reflect the inherent economic
nature of Saudi Arabia, which is predominantly an oil-dominated economy. Second, they
could be indicative of relative under-development of the Saudi banking system, which could
lead to imbalances between saving and investment and may distort investment decisions. This
is in line with Malkawi et al. (2012), who argue that the financial sector in Saudi Arabia is
still in the transition stage. Hence, it needs to go beyond a certain threshold before it can be
instrumental in promoting economic growth.
These findings also highlight the specific nature of oil and resource-rich economies like
Saudi Arabia. Resource-driven economies do not necessarily follow the same patterns of
development as manufacturing economies. The economy crucially depends on price
fluctuations and foreign markets, as documented by the strong role played in our analysis by
the oil price and openness to trade. Financial development does not play as prominent a role
as in industrialized economies, or may not even play any role at all. The two arguments
mentioned in the preceding paragraph may therefore be related: the fact that the Saudi
banking sector is underdeveloped may itself be due to the dominant role of oil in the
economy. Banking plays an important role in industrialized and agricultural economies alike,
in that it improves allocation of resources to firms and helps these firms stay afloat until their
goods are sold. This role is less important when the economy is dominated by extraction of a
highly liquid (in financial sense) and easily marketable commodity.
22
Our results suggest, nevertheless, that the Saudi non-oil sector is favourably affected by
financial development. Hence, from a policy perspective, it is useful to further develop the
Saudi banking system with a view to aiding the growth of the non-oil sector, given that the
impact of financial development on the latter is positive and significant. In that way, and if
the diversification of the Saudi economy continues, we can anticipate that financial
development will play a more prominent role in the country’s overall economic performance
in the future.
23
References
Atje, Raymond and Boyan Jovanovic. (1993). Stock Markets and Development. European
Economic Review, 37, 632-640.
Abu-Bader, S., and Abu-Qarn, A.S. (2008). Financial Development and Economic Growth:
The Egyptian Experience. Journal of Policy Modeling, 30 (5), 887–898.
Albatel, A.H. (2000). Financial Development and Economic Growth in Saudi Arabia. Journal
of Economic & Administrative Sciences, 16, 162-188.
Al-Hassan, A., Khamis,M. and Oulidi.N. (2010). The GCC Banking Sector: Topography and
Analysis, International Monetary Fund, IMF Working Paper, 10-87.Al-Malkawi, H., and
Abdullah, N. (2012). Finance-Growth Nexus: Evidence from a Panel of MENA
Countries. International Journal of Economics and Finance, 4 (5), 129-139.
Al-Yousif, Y. K. (2002). Financial Development and Economic Growth: Another Look at the
Evidence from Developing Countries. Review of Financial Economics, 11(2), 131–150.
Ang, J. B., and McKibbin,W.J. (2007). Financial development and Economic Growth:
Evidence from Malaysia. Journal of Development Economics, 84, 215-233.
Ang,James B ,(2010) financial development and economic growth in Malaysia. Routledge
Studies in the Growth Economies of Asia.
Agung, F. and Ford, J. (1998). Financial Development, Liberalization and Economic
Development in Indonesia, 1966-1996: Cointegration and Causality. University Of
Birmingham, Department of Economics Discussion Paper No: 98-12.
Asterious, Dimitrios and Stephen Hall, 2011, Applied Econometrics: A Modern Approach,
Palgrave Macmillan. Second edition.
Balajas, A., Chami, R., and Yousefi, R.S. (2011). The Finance-Growth Nexus Re-Examined:
Are There Cross-Region Differences? IMF Working Paper, 48(3).
Beck, Thorsten. (2010). Finance and Oil: Is There a Resource Curse in Financial
Development?. European Banking Centre Discussion Paper No. 2011-004. (Tilburg, The
Netherlands: Tilburg University).
Berthelemy, J.C., and Varoudakis, A. (1997). Economic Growth, Convergence Clubs, and the
Role of Financial Development. Oxford Economic Papers 48, 300–328.
24
Buffie, E. F. (1984). Financial Repression, the New Structuralists, and Stabilization Policy in
Semi-Industrialized Economics, Journal of Development Economics, 14, 305-22.
Choudhury, M.A and Al-Sahlawi, M. A. (2000).Oil and Non-oil Sectors in the Saudi Arabian
Economy. OPEC Review, 24 (3), 235–250.
Christopoulos, D. K. and Tsionas, E. G. (2004) ‘Financial development and economic
growth: Evidence from panel unit root and cointegration tests’, Journal of Development
Economics, 73, 55-74.
Demirguc-Kunt, Asli and Vojislav Maksimovic. (1998). Law, finance and firm growth,
Journal of Finance, 53(6), 2107-2137.
Dara, L. and Sovannroeun, (2008) “The Monetary Model of Exchange Rate: Evidence from
the Phillipines using ARDL Approach”, Munich Personal RePEc Archive (MPRA).
De Gregorio, J., and Guidotti, P. (1995). Financial Development and Economic Growth.
World Development, 23, 433-448.
Demetriades, P. and Law, S.H. (2006). Finance, Institutions and Economic
Performance. International Journal of Finance and Economics, 11, 1 – 16.
Demetriades, P., and Hussein, K. (1996). Does Financial Development Cause Economic
Growth? Time Series Evidence from Sixteen Countries. Journal of Development
Economics, 51(2), 387–411.
Doing Business in Saudi Arabia 2012. World Bank. Available at
http://www.doingbusiness.org/data/exploreeconomies/saudi-arabia/. (Accessed 22
November 2011).
Durlauf, Steven, Paul Johnson, and Jonathan Temple, “Growth Econometrics” (pp. 557–677),
in Philippe Aghion and Steven Durlauf (Eds.), Handbook of Economic Growth
(Amsterdam: North-Holland, 2005).
Fry, M.J. (1978). Money and capital or financial deepening in economic development.
Journal of Money, Credit and Banking 10, 464–475.
Galbis, V., (1977). Financial Intermediation and Economic Growth in Less-Developed
Countries: A Theoretical Approach. Journal of Development Studies, 13(2), 58-72.
Ghatak, S. (1997) Financial Liberalisation: The Case of Sri Lanka. EmpiricalEconomics,
22(1), 117–131.
25
Greenwood, J., Bruce, S. (1997). Financial Markets in Development, and the Development of
Financial Markets. Journal of Economic Dynamic and Control, 21(1), 145–181.
Greenwood, J., Jovanovic, B. (1990). Financial Development, Growth and the Distribution of
Income. Journal of Political Economy, 98 (5), 1076–1107.
Hondroyiannis, G., Sarantis, L., and Evangelia, P. (2005). Financial Market and Economic
Growth in Greece, 1986-1999. Journal of International Financial Market, Institution and
Money, (15),173-188.
Hosein, S. (2007) ‘Demand for money in Iran: An ARDL approach’, Department of
Economics, Islamic Azad University, Khorasgan Branch, Isfahan, Iran.
Hicks, J. (1969). A Theory of Economic History. Clarendon Press, Oxford.
Hendry, D. F. (1995a) Dynamic Econometrics, Oxford University Press, Oxford.
Inder, B. (1993). Estimating Long-Run Relationship in Economics: A Comparison of
Different Approaches. Journal of Econometrics, 57: 53-68.
Jalil, A., and Ma, Y. (2008). Financial Development and Economic Growth: Time Series
Evidence from Pakistan and China. Journal of Economic Cooperation, 29(2), 29-68.
King, R. G. and Levine, R. (1993). Finance and growth: Schumpeter Might Be Right.
Quarterly Journal Economics, 108: 717-37.
Kenny, C., and Williams, D. (2001). What Do We Know about Economic Growth? Or, Why
Don’t We Know Very Much? World Development, 29(1), 1–22.
Khan, M. A., Qayyum, A. and Saeed, A. S. (2005). Financial Development and Economic
Growth: The Case of Pakistan”, The Pakistan Development Review 4. 44, 819-837.
Krugman, P.R. (1979). Increasing Returns, Monopolistic Competition and International
Trade. Journal of International Economics, 9 (4), 469-479.
Krugman, P.R. (1980). Scale Economies, Product Differentiation, and the Pattern of Trade.
American Economic Review, 70 (5), 950-959.
Kuznets, S. (1955). Economic Growth and Income Inequality. The American Economic
Review, 45(1), 1–28.
Levine, Ross and Zervos, Sara (1998), “Stock Markets, Banks, and Growth,” American
Economic Review, 88(3), 537-558.
26
Levine, R., and Beck, T. (1999). A new database on financial development and
structure. Research Working papers, 1(1), 1-69.
Levine, R. and Renelt, D. (1992), “A Sensitivity Analysis of Cross-Country
GrowthRegressions,” American Economic Review 82 (4), 942-963.
Lyons, S.E. and Murinde, V. (1994) .Cointegration and Granger-Causality Testing of
Hypotheses on Supply-Leading and Demand-Following Finance. Economic Notes. 23
(2), 308-316.
Loayza, N. V., and Ranciere, R. (2006). Financial Development, Financial Fragility, and
Growth. Journal of Money, Credit and Banking. 38(4), 1051-1076.
Lucas, R. (1988). On the Mechanics of Economic Development. Journal of Monetary
Economics 22, 2-42.
Luintel, K. B., Khan, M. 1999. A quantitative reassessment of the finance-growth nexus,
Evidence from a multivariate VAR, Journal of Development Economics, 60, 381-405.
Majid, M. S. (2007). Does Financial Development and Inflation Spur Economic Growth in
Thailand? Chulalongkorn Journal of Economics, 19, 161–184.
Mahran, H. A. (2012). Financial Intermediation and Economic Growth in Saudi Arabia: An
Empirical Analysis, 1968-2010. Modern Economy, 3.
McKinnon R. (1973) Money and Capital in Economic Development (Washington: The
Brookings Institute).
Mohamad, S. E. (2008). Finance-Growth Nexus In Sudan: Empirical Assessment Based on
an Application of the Autoregressive Distributed Lag (ARDL) Model. Working Paper,
No.API/WPS 0803, Arab Planning Institute, Kuwait.
Murinde, V., and Eng, F.S.H. (1994). Financial Development and Economic Growth in
Singapore: Demand-Following or Supply-Following. Applied Financial Economics, 4
(6), 391-404.
Muhsin, K. and Eric, J. P. (2000). Financial Development and Economic Growth in Turkey:
Further Evidence on the Causality Issue. Centre for International, Financial and
Economics Research Department of Economics Loughborough University.
27
Nili, M. and M. Rastad. (2007). Addressing the Growth Failure of the Oil Economies: The
Role of Financial Development. The Quarterly Journal of Economics and Finance, 46,
726–40.
Odedokun, M.O. (1989). Causalities between Financial Aggregates and Economic Activities:
The Results from Granger’s Test. Savings and Development, 23 (1), 101- 111.
Patrick, H.T. (1966). Financial Development and Economic Growth in Underdeveloped
Countries. Economic Development and Cultural Change, 14(2), 174-189.
Pesaran, H. M. (1997). The Role of Economic Theory in Modelling the Long-Run. Economic
Journal, 107, 178-191.
Pesaran, M. H., Shin, Y., and Smith, R. J. (2001). Bounds Testing Approaches to the
Analysis of Level Relationships. Journal of Applied Econometrics, 16, 289-326.
Ram, R. (1999). Financial Development and Economic Growth: Additional Evidence. The
Journal of Development Studies, 35 (4), 164-174.
Qi, L. and Teng, J. Z. (2006). Financial Development and Economic Growth: Evidence from
China. China Economic Review, 17(3), 395-411.
Robinson, J. (1952). The Generalization of the General Theory. In The Rate of Interest, and
other Essays. London: McMillan, 67–146.
Safdari, M., Mehrizi, M.A., and Elahi, M. (2011). Financial Development and Economic
Growth in Iran. European Journal of Economics, Finance and Administrative Sciences,
36, 115-122.
Singh, T. (2008). Financial Development and Economic Growth Nexus: A Time-Series
Evidence from India. Applied Economics 40, 1615-1627.
Schumpeter, J. A. (1934). The Theory of Economic Development. Translated by Redvers
Opie, Cambridge MA: Harvard University Press.
Shaw, E. S. (1973). Financial Deepening in Economic Development. Oxford University
Press, New York.
Shrestha, M.B. and Chowdhury, K. (2005). ARDL Modelling Approach to Testing the
Financial Liberalization Hypothesis. Economics Working Paper Series 2005, University
of Wollongong.
28
Stern, N. (1989). The Economics of Development: A Survey. The Economic Journal,
99(397), 597-685.
Thangavelu, S.M. and Ang, B. J. (2004). Financial Development and Economic Growth in
Australia: An Empirical Analysis, Empirical Economics, 29, 247-260.
Thakor, A.V. (1996). The design of financial systems: An Overview. Journal of Banking and
Finance, 20, 917-948.
The Political Risk Services Country Reports (PRS )Group Saudi Arabia Country Forecast,
May 2010 The Saudi Insurance Market Report 2009.
van der Ploeg, F. (2011). Natural Resources: Curse of Blessing? Journal of Economic
Literature, 29 (2), 366-420.
Van Wijnbergen. S. (1983). Credit Policy, Inflation and Growth in a Financially Repressed
Economy. Journal of Development Economics, 3(1-2), 45-65.
Woo, J. (2009). Why Do More Polarized Countries Run More Procyclical Fiscal Policy? The
Review of Economics and Statistics, 91(4), 850-870.
Wood A. (1993). Financial Development and Economic Growth in Barbados: Causal
Evidence. Savings and Development, 17 (4), 379-389.
Wu, J., Hou, H., and Cheng, S. (2010). The Dynamic Impacts of Financial Institutions on
Economic Growth: Evidence from the European Union. Journal of Macroeconomics, 32,
879-891.
Xu, Z. (2000). Financial Development, Investment and Growth. Economic Inquiry 38(2),
331-344.
29
Tables
Table 1: Selected Indicators of Banking Sector
Year M3/GDP PRIVATE/GDP Total Bank Asset
2005 46.8218 36.8644469 759075
2006 49.4604 35.64138057 861088
2007 54.7463 40.05913986 1075221
2008 52.0185 41.12532216 1302271
2009 72.8406 52.53976349 1370258
2010 64.3419 47.59243453 1415267
Sources: SAMA 48th Annual Report.
Table 2: Bank Credit to the Private Sector by economic Activity (In Million Riyals)
2006 2007 2008 2009 2010
Amount % Share
Amount % Share
Amount % Share
Amount % Share
Amount % Share
Agriculture & Fishing
6802 1.5 8636 1.5 10980 1.5 8731 1.2 10269 1.4
Manufacturing & Processing
37566 8.1 54339 9.7 79333 11.1 75044 10.6 90082 12.1
Mining & Qurrying
1802 0.4 3897 0.7 4265 0.6 5337 0.8 5818 0.8
Electricity, Water & Gas
3598 0.8 5878 1.1 10629 1.5 13365 1.9 19243 2.6
Building & Construction
37845 8.2 43421 7.8 54371 7.6 44741 6.3 55644 7.5
Commerce 111511 24.1 127473 22.9 176858 24.8 169220 23.9 181132 24.4
Transport & Communication
6875 1.5 20989 3.8 37814 5.3 38415 5.4 42992 5.8
Finance 61828 13.4 62632 11.2 16812 2.4 21258 3.0 17756 2.4
Services 16735 3.6 28286 5.1 32324 4.5 46123 6.5 35660 4.8
Miscellaneous 177539 38.4 201854 36.2 289351 40.6 286536 40.4 284461 38.3
Total 462,103 100 557,405 100 712,737 100 708,769 100 743,057 100 Sources: SAMA 47th Annual Report.
30
Table 3: Principal Components Analysis
Number of Obs = 41 Number of comp. = 3
Component Eigenvalue Difference Proportion Cumulative
Comp1 2.912 2.840 0.971 0.971
Comp2 0.072 0.0569 0.024 0.995
Comp3 0.015 . 0.005 1.000
Table 4: Unitroot Test Variables ADF test ADF test
In level I(0) First difference I(1)
Intercept Intercept & trend Intercept Intercept &trend
GDP -2.598 -3.078* -2.997** -3.463*
GDPN -3.15** -3.371* -2.47 -2.82
GDPO -2.659* -3.450* -5.335*** -5.394***
FD -0.250 -2.621 -6.999*** -7.004***
OILP -2.631* -2.401 -6.028*** -6.022***
TRD -1.555 -1.491 -9.097*** -9.001*** Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 5: Result from Bound test Dep. Var. SIC Lag F-statistic Probability Outcome
FGDP(GDP|FD, OILP, TRD) 1 7.580 0.000*** Cointegration
FFD(FD|GDP, OILP, TRD) 1 3.636 0.015** Cointegration
FOILP(OILP| FD, GDP, TRD) 1 3.355 0.021** Cointegration
FTRD(TRD| FD, GDP, OILP) 1 1.254 0.308 No Cointegration
Notes: Asymptotic critical value bounds are obtained from Table F-Statistic in appendix CI, Case II: intercept and no trend for k=4 (Pesaran et al (2001) p.300).
31
Table 6: Estimated Long Run Coefficients using the ARDL Approach ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion, Dependent variable is GDP
Regressor Coefficient Standard Error T-Ratio Probability
C -6.950 12.390 -0.560 0.579
FD -0.033 0.035 -0.962 0.342
OILP 0.133*** 0.023 5.690 0.000
TRD 2.14*** 0.088 24.310 0.000
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 7: Error Correction Representation for the Selected ARDL Model ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion. Dependent variable is ΔGDP
Regressor Coefficient Standard Error T-Ratio Probability
C 1.750** 0.805 2.173 0.037
ΔFD -0.004 0.004 -0.993 0.327
ΔOILP 0.001 0.004 0.252 0.802
ΔTRD 0.118* 0.058 1.74 0.089
ecm(-1) -0.128*** 0.023 -5.47 0.000
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 8: ARDL-VECM Model Diagnostic Tests R2=0.96, Adjusted R2=0.95
Serial Correlation (1)=0.001[0.972] Normality (2)=1.687[0.43] Functional Form (1)= 0.559[0.454] Heteroscedasticy (1)=1.640[0.199] Figure 1: Plot of Cusum and Cusumq for coefficients stability for ECM model (1)
Plot of Cumulative Sum of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-5 -10 -15 -20
0 5
10 15 20
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
Plot of Cumulative Sum of Squares of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-0.5
0.0
0.5
1.0
1.5
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
32
Table 9: Result from Bound test Dep. Var. SIC Lag F-statistic Probability Outcome
FGDPN (GDPN| FD, OILP, TRD) 2 10.381 0.000*** Cointegration
FFD (FD| GDPN, OILP, TRD) 1 4.199 0.007** Cointegration
FOILP(OILP| FD, GDPN, TRD) 1 5.996 0.001** Cointegration
FTRD(TRD| FD, GDPN, OILP) 1 2.770 0.042* Cointegration
Notes: Asymptotic critical value bounds are obtained from Table F-Statistic in appendix CI, Case II: intercept and no trend for k=4 (Pesaran et al (2001) p.300). Table 10: Estimated Long Run Coefficients using the ARDL Approach ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion, Dependent variable is GDPN
Regressor Coefficient Standard Error T-Ratio Probability
C 1.25** 0.600 2.070 0.040
FD 0.184* 0.106 1.730 0.091
OILP 0.078 0.046 1.660 0.104
TRD 2.14*** 0.088 24.310 0.000
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 11: Error Correction Representation for the Selected ARDL Model ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion. Dependent variable is ΔGDPN
Regressor Coefficient Standard Error T-Ratio Probability
C 1.918*** 0.702 2.729 0.010
ΔFD 0.111 0.008 1.390 0.172
ΔOILP 0.110*** 0.004 2.570 0.014
ΔTRD 0.061 0.062 0.980 0.333
ecm(-1) -0.06*** 0.174 -3.450 0.001
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 12: ARDL-VECM Model Diagnostic Tests R2=0.99, Adjusted R2=0.98
Serial Correlation (1)=.010[0.91] Normality (2)=0.053[0.97] Functional Form (1)= .016[0.89] Heteroscedasticy (1)=4.65[0.031]
33
Figure 2: Plot of Cusum and Cusumq for coefficients stability for ECM model (2)
Table 13: Result from Bound test Dep. Var. SIC Lag F-statistic Probability Outcome
FGDPO(GDPO|FD, OILP, TRD) 1 3.840 0.017** Cointegration
FFD(FD|GDPO, OILP, TRD) 1 1.313 0.297 No Cointegration
FOILP(OILP| FD, GDPO, TRD) 1 2.504 0.068 Inconclusive
FTRD(TRD| FD, GDPO, OILP) 1 1.959 0.138 No Cointegration
Notes: Asymptotic critical value bounds are obtained from Table F-Statistic in appendix CI, Case II: intercept and no trend for k=4 (Pesaran et al (2001) p.300). Table 14: Estimated Long Run Coefficients using the ARDL Approach ARDL(1,1,0,0) selected based on Schwarz Bayesian Criterion, Dependent variable is GDPO
Regressor Coefficient Standard Error T-Ratio Probability
C 4.100 6.060 .676 0.504
FD 0.170 .123 1.44 0.157
OILP 0.193** .082 2.35 0.025
TRD 3.140*** .158 19.87 0.000
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 15: Error Correction Representation for the Selected ARDL Model ARDL (1,1,0,0) selected based on Schwarz Bayesian Criterion. Dependent variable is ΔGDPO
Regressor Coefficient Standard Error T-Ratio Probability
C 3.584** 1.744 2.054 0.048
ΔFD -0.088** 0.044 -2.004 0.053
ΔOILP 0.021*** 0.007 2.954 0.006
ΔTRD 0.349** 0.149 2.340 0.025
ecm(-1) -0.111** 0.051 -2.155 0.038
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1.
Plot of Cumulative Sum of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-5 -10 -15 -20
0 5
10 15 20
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
Plot of Cumulative Sum of Squares of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-0.5
0.0
0.5
1.0
1.5
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
34
Table 16: ARDL-VECM model diagnostic tests R2=0.77, Adjusted R2=0.73
Serial Correlation (1)=2.049[0.152] Normality (2)=.0211[0.989] Functional Form (1)= 2.291[0.130] Heteroscedasticy (1)=14.860[0.00] Figure 3: Plot of Cusum and Cusumq for coefficients stability for ECM model (3)
Robustness check tables results: Table 17: Principal Components Analysis Number of Obs = 41 Number of comp. = 3
Component Eigenvalue Difference Proportion Cumulative Comp1 2.907 2.853 0.969 0.969 Comp2 0.539 0.015 0.018 0.987 Comp3 0.038 . 0.012 1.000
Table 18: Results from Bound test
Dep. Var. SIC Lag F-stat. Probability Outcome
FGDP(GDP|FD2, OILP, TRD) 1 6.763 0.000*** Cointegration
FFD2(FD2|GDP, OILP, TRD) 1 1.825 0.148 No Cointegration
FOILP(OILP| FD2,GDP, TRD) 1 3.861 0.011** Cointegration
FTRD(TRD|FD2, GDP, OILP) 1 2.924 0.304 Inconclusive
Notes: Asymptotic critical value bounds are obtained from Table F-Statistic in appendix CI, Case II: intercept and no trend for k=4 (Pesaran et al (2001) p.300).
Plot of Cumulative Sum of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-5 -10 -15 -20
0 5
10 15 20
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
Plot of Cumulative Sum of Squares of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-0.5
0.0
0.5
1.0
1.5
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
35
Table 19: Estimated Long Run Coefficients using the ARDL Approach ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion, Dependent variable is GDP
Regressor Coefficient Standard Error T-Ratio Probability
C 4.588 3.334 1.375 0.178
FD2 -0.399** 0.172 - 2.313 0.027
OILP 0.053** 0.023 2.326 0.026
TRD 0.028** 0.013 2.205 0.035
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 20: Error Correction Representation for the Selected ARDL Model ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion. Dependent variable is ΔGDP
Regressor Coefficient Standard Error T-Ratio Probability
C 0.752 0.762 0.986 0.331
ΔFD2 - 0.094 * 0.052 -1.796 0.082
ΔOILP 0.007** 0.003 1.994 0.054
ΔTRD 0.004** 0.002 2.305 0.027
ecm(-1) -0.164*** 0.053 -3.085 0.004
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 21: ARDL-VECM Model Diagnostic Tests R2 =0.97, Adjusted R2 =0.96
A:Serial Correlation (1)= 0.128[0.720] C:Normality (2)= 0.894[0.639]
B:Functional Form (1)= 2.526[0.112] D:Heteroscedasticity (1)= 0.135[0.712]
Figure 4: Plot of Cusum and Cusumq for coefficients stability for ECM- Robustness model (1)
Plot of Cumulative Sum of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-5 -10 -15 -20
0 5
10 15 20
1970 1975 1980 1985 1990 1995 2000 2005 20102010
Plot of Cumulative Sum of Squares of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-0.5
0.0
0.5
1.0
1.5
1970 1975 1980 1985 1990 1995 2000 2005 20102010
36
Table 22: Results of Bound test
Dep. Var. SIC LaG F-stat. Probability Outcome
FGDPN(GDPN|FD2, OILP, TRD) 2 7.4093 0.001*** Cointegration
FFD(FD2|GDPN, OILP, TRD) 2 3.084 0.030 No Cointegration
FOILP(OILP|GDPN, FD2, TRD) 2 3.322 0.022 No Cointegration
FTRD(TRD|GDPN, FD2, OILP) 2 5.835 0.001*** Cointegration
Notes: Asymptotic critical value bounds are obtained from Table F-Statistic in appendix CI, Case II: intercept and no trend for k=4 (Pesaran et al (2001) p.300).
Table 23: Estimated Long Run Coefficients using the ARDL Approach ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion, Dependent variable is GDPN
Regressor Coefficient Standard Error T-Ratio Probability
C 0.492*** 0.122 4.049 0.000
FD2 0.014* 0.007 1.879 0.077
OILP 0.010 0.082 0.121 0.904
TRD 0.015*** 0.003 4.592 0.000
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1.
Table 24: Error Correction Representation for the Selected ARDL Model ARDL(2,0,1,1) selected based on Schwarz Bayesian Criterion. Dependent variable is ΔGDP
Regressor Coefficient Standard Error T-Ratio Probability
C 0.535 0.577 0.928 0.359
ΔFD2 0.106*** 0.024 4.296 0.000
ΔOILP 0.101 * .056973 1.7901 0.082
ΔTRD 0.010*** 0.002 3.897 0.000
ecm(-1) -0.066* 0.037 -1.768 0.086
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1.
37
Table 25: ARDL-VECM Model Diagnostic Tests R2 =0.99, Adjusted R2 =0.98
Serial Correlation (1)= 0.454 [0.50] Normality ( 2)= 0.972[0.97]
Functional Form (1)= 0.972 [0.61] Heteroscedasticity (1)= 3.203[0.07]
Figure 5: Plot of Cusum and Cusumq for coefficients stability for ECM- Robustness model (2)
Table 26: Results of Bound test
Dep. Var. SIC Lag F-stat. Probability Outcome
FGDPO(GDPO|FD2, OILP, TRD) 2 4.837 0.007** Cointegration
FFD2(FD2|GDPO, OILP, TRD) 2 2.266 0.084 No Cointegration
FOILP(OILP|GDPO, FD2, TRD) 2 3.467 0.018 No Cointegration
FTRD(TRD|GDPO, FD2, OILP) 2 0.764 0.556 No Cointegration
Notes: Asymptotic critical value bounds are obtained from Table F-Statistic in appendix CI, Case II: intercept and no trend for k=4 (Pesaran et al (2001) p.300).
Table 27: Estimated Long Run Coefficients using the ARDL Approach ARDL(1,1,0,0) selected based on Schwarz Bayesian Criterion, Dependent variable is GDPO
Regressor Coefficient Standard Error T-Ratio Probability
C 9.587*** 1.882 5.093 0.000
FD2 -0.435*** 0.128 -3.400 0.002
OILP 0.053** 0.028 1.875 0.069
TRD 0.060*** 0.014 4.172 0.000
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1.
Plot of Cumulative Sum of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-5 -10 -15 -20
0 5
10 15 20
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
Plot of Cumulative Sum of Squares of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-0.5
0.0
0.5
1.0
1.5
1969 1974 1979 1984 1989 1994 1999 2004 2009 2010
38
Table 28: Error Correction Representation for the Selected ARDL Model ARDL (1,1,0,0) selected based on Schwarz Bayesian Criterion. Dependent variable is ΔGDPO
Regressor Coefficient Standard Error T-Ratio Probability
C 3.515** 1.411 2.490 0.018
ΔFD2 -0.159*** 0.057 -2.770 0.009
ΔOILP 0.019** 0.007 2.584 0.014
ΔTRD 0.022*** 0.007 3.086 0.004
ecm(-1) -0.366*** 0.106 -3.455 0.001
Note:*, **, and *** indicate significance at* 10 %, ** at 5 % and *** at 1. Table 29: ARDL-VECM Model Diagnostic Tests R2 =0.84, Adjusted R2 =0.81
Serial Correlation (1)= 0.638[1.00] Normality (2)= 0.233[0.890]
Functional Form (1)= 0.130[0.718] Heteroscedasticity (1)= 7.605[0.006]
Figure 6: Plot of Cusum and Cusumq for coefficients stability for ECM- Robustness model (3)
Plot of Cumulative Sum of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-5 -10 -15 -20
0 5
10 15 20
1970 1975 1980 1985 1990 1995 2000 2005 20102010
Plot of Cumulative Sum of Squares of Recursive Residuals
The straight lines represent critical bounds at 5% significance level
-0.5
0.0
0.5
1.0
1.5
1970 1975 1980 1985 1990 1995 2000 2005 20102010
View publication statsView publication stats
