The current issue and full text archive of this journal is available at www.emeraldinsight.com/1086-7376.htm SEF 28,2 Saving and investment in Saudi Arabia: an empirical analysis 136 Reetu Verma School of Economics, Faculty of Commerce, University of Wollongong, Wollongong, Australia, and Ali Salman Saleh Accounting, Economics, Finance & Law (AEFL) Group, Faculty of Business and Enterprise, Swinburne University of Technology, Hawthorn, Australia Abstract Purpose – This paper examines the long-term relationship between saving and investment as a criterion for assessing international capital mobility for the case of Saudi Arabia, the largest economy among the Middle Eastern and Arab nations. Design/methodology/approach – The approach is modeled on Feldstein and Horioka covering the period 1963-2007 for Saudi Arabia. We use the bounds testing approach and the Gregory and Hansen cointegration methods to test for the long-run relationship between saving and investment. Additionally, before testing for this relationship, we conduct unit root tests, including the additive outlier model developed by Perron with an endogenously determined structural break. Findings – The study finds no evidence of a long-run relationship between saving and investment and therefore concludes that capital is highly mobile in Saudi Arabia. This finding is plausible given the economic and financial reforms which have occurred in Saudi Arabia along with increased capital inflows into the country in the last few decades. Originality/value – Of the limited studies so far on developing countries, none has considered the capital mobility issue for an oil-dependent country. Keywords Saving, Investment, Capital mobility, Saudi Arabia Paper type Research paper Studies in Economics and Finance Vol. 28 No. 2, 2011 pp. 136-148 q Emerald Group Publishing Limited 1086-7376 DOI 10.1108/10867371111137139 1. Introduction A sound grasp of the relationship between saving and investment is important for understanding a country’s growth and economic development. Further to this, growth models suggest that it is the amount of capital accumulation that determines the growth of output and the amount of capital accumulation in an economy is ultimately constrained by its saving rate. As the economy increases its saving, more funds will be available for investment. Thus, the issue of correlation between saving and investment assumed importance in economics. Much of the literature in this area focuses on the relationship between saving and investment at both theoretical and empirical levels. In their seminal paper, Feldstein and Horioka (1980) interpret the high correlation between saving and investment rates as evidence of low international capital mobility. Theoretically, in a closed economy with low capital mobility, domestic saving finances JEL classification – C22, E20, E30 all investment and hence the correlation is high. However, in an open economy, domestic saving is not necessarily used to finance domestic investment, but will be used to gain better returns in international capital markets. Contrary to these expectations, Feldstein and Horioka (1980) find that capital is not very mobile internationally among developed countries. Since then, many studies examine the relationship between saving and investment for different time periods, data sets and country samples; both time-series and cross-section studies exist. While the Feldstein and Horioka’s (1980) proposition emphasizes the empirical association between savings and investment, no consensus from the literature has emerged explaining this link. Using US data from 1946 to 1987, Miller (1988) studies the relationship between saving and investment by using cointegration techniques. He finds that the two variables shared a cointegrating relationship prior to the Second World War period but the long-run relationship did not exist after that period, leading Miller to conclude that the findings could be explained by the increased level of international capital mobility after the war. Many studies investigate the observed correlation between domestic saving and investment in the European Union (EU) countries including Arginon and Roldan (1994), Apergis and Tsoulfidis (1997), and Kollias et al. (2008). The first two studies find that saving and investment are cointegrated and therefore conclude that capital is not very mobile in the EU countries. Kollias et al. (2008) examine the saving-investment correlation using the autoregressive distributed lag (ARDL) approach and panel regressions for 15 EU member countries from 1962 to 2002. They find that a long-run relationship between the two variables exists only for eight countries[1]. The authors accept the Feldstein and Horioka explanation and interpret the finding as evidence of high capital mobility. Using data covering the 1975-1995 period and dividing 90 developing countries into four regions of Africa, Asia, Latin America, and the Middle East. Isaksson (2001) finds that capital is relatively immobile for developing countries. On the other hand, AbuAl-Foul (2006) examines the relationship between saving and investment rates by using the Johansen and Juselius (1990) cointegration procedure. He finds that the two variables are not cointegrated and concludes that capital is mobile in the four MENA countries[2]. Ho (2000) extends the Feldstein and Horioka model to test for the capital mobility issue by drawing two samples from two different regimes for Taiwan. He concludes that the Feldstein and Horioka model is supported for the more open regime. Jansen (1996) uses an error correction model to study the saving-investment relationship and finds that the two rates have a long-run relationship for most of the Organisation for Economic Co-operation and Development (OECD) countries. This is different from Chaudhri and Wilson (2000) and Sachsida and Mendonça (2006) who find no long-run relationship between saving and investment for Australia and Brazil, respectively. Cooray and Sinha (2007) study the relationship between saving and investment rates for 20 African countries. They use both the Johansen cointegration and fractional cointegration tests which indicate mixed results. The Johansen cointegration test shows that the saving and investment rates are cointegrated only for Rwanda and South Africa, implying that for the other 18 countries, there is evidence of capital mobility. However, the two rates are found to be fractionally cointegrated in 12 of the 20 countries examined. Last, Ang (2007) examines the cointegrating relationship between domestic saving and investment in Malaysia using the ARDL framework. Using data for the period 1965-2003, he finds a long-run relationship between domestic saving and investment in Malaysia. Saving and investment 137 SEF 28,2 138 There is one strand of the theoretical literature which departs from the FeldsteinHorioka approach. These studies describe a strong saving-investment correlation in the presence of high capital mobility. They argue that the saving-investment correlation is due to other macroeconomic factors such as country size (Baxter and Crucini, 1993), non-traded goods (Murphy, 1986; Wong, 1990), current account solvency (Coakley et al., 1996) and financial structure (Kasuga, 2004). But even here the empirical results resulting from these studies vary considerably. The objective of this study is to investigate whether a long-run relationship exists between saving and investment in Saudi Arabia as a criterion for assessing the capital mobility issue for this country. Our paper contributes to the existing literature in three ways. First, of the limited studies so far on developing countries, none has considered the issue of capital mobility for an oil-dependent country. Saudi Arabia is an interesting case: not only is it is the largest economy among the Middle Eastern and Arab nations, but also its economy is dominated by the oil sector. The oil sector accounts for 75-85 percent of government revenue and 90 percent of export revenue (Economic Intelligence Unit (EIU, 2007). The country has also undergone remarkable changes in its financial sector. The changes include the establishment of credit agencies and banks during the 1960s and 1970s and the stock market in the 1980s leading to a considerable financial and monetary system revolution (Albatel, 2003). Second, studies on the saving-investment correlation fail to take into account any potential structural break in their stationarity and/or cointegration estimations. It is well known that traditional tests which do not allow for a structural break may produce spurious results. Last, the availability of lengthy data series for Saudi Arabia also makes it an eminently suitable case as it allows for a robust, in-depth country analysis to be conducted. According to Ang (2007, p. 2168), “cross-sectional empirical analyzes conducted at the aggregate level are unable to capture and account for the complexity of financial environments and economic histories of each individual country”. The aim of this paper is achieved in two-steps. In the first step, a detailed unit root treatment of the data series is undertaken to establish their order of integration. The idea behind this exercise is to ascertain whether, in the presence of structural break in the data, the series are integrated of order one, or otherwise. To accomplish this step, Perron’s (1997) one break unit root test with an endogenously determined structural break is used. In the second step, the ordinary least square-based ARDL bounds test (Pesaran et al., 2001) and the Gregory and Hansen (1996) methods for cointegration between Saudi Arabia’s saving and investment are applied. The remainder of the paper is organized as follows. Section 2 overviews the economy of Saudi Arabia and Section 3 examines the conceptual model of the study. Sections 4 and 5 discuss the unit root and cointegration methodologies. The data and empirical results are reported in Section 6. Section 7 concludes with some policy implications. 2. An overview: Saudi Arabia Saudi Arabia is a very wealthy, oil producing country in the Middle East and has the largest economy in the Arab and Gulf region. According to Al-Rajhi et al. (2003), gross domestic product (GDP) in Saudi Arabia has increased dramatically since 2001 such that Saudi Arabia is now by far the largest economy in the region, almost twice the size of Israel or Egypt. However, the country’s growth is highly volatile due to its dependency on oil revenues which are prone to price fluctuations. High economic growth occurred in Saudi Arabia during the 1970s and early 1980s but the country suffered a recession in the mid 1980s. Saudi Arabia’s economy subsequently strengthened and again experienced strong growth in the late 1980s and early 1990s. During the period 1993-2002, real GDP growth averaged at approximately 1.5 percent a year. The rate of real GDP growth increased to 7.7 percent per annum and the growth rate of nominal GDP rose to 19.8 percent. These growth rates were due to an increase in oil revenues as well as higher oil prices in the overseas market. The high oil prices during 2004-2006 contributed to an average real GDP growth of approximately 5 percent and an average nominal GDP of approximately 17 percent (EIU, 2007). Despite the government’s attempt to diversify its economy to non-oil sectors and to promote private sector businesses, oil remains the major source of government revenue. This sector accounted for 50 percent of GDP during 2005-2006 and provided 75-85 percent of government revenue (EIU, 2007). These figures were driven mainly by oil exports and the accumulation of domestic capital. More recently, the non-oil sector has attracted attention as the government is starting to realize the importance of diversifying the economy. According to EIU (2007), 4.3 percent of real GDP growth for the year 2006 was attributable to non-oil sectors and the private sector played an important role in these activities. As shown in Figure 1, the country achieved remarkably high saving rates during the 1970s and early 1980s indicating that Saudi Arabia was successful in mobilizing saving. However, with expanded investment in infrastructure, the share of domestic saving as a percentage of GDP decreased from approximately 60 percent in 1970 to about 25 percent in 1992. The dramatic fall was mainly due to the Gulf War and the energy crisis of the early 1990s. Domestic saving once again started to increase during the late 1990s and 2000s reaching approximately 50 percent of GDP in 2007. This result was achieved once again as a result of the improvement in oil prices, producing larger oil revenues and increase in saving. Along with the rise in saving, broad money in the country increased during 2006-2007; for example, broad money based on 10 months’ data grew from 11.8 percent in 2006 to over 13 percent in 2007 (SAMBA, 2008)[3]. Gross domestic investment also increased from an average of 11 percent in the 1970s to 20 percent in 1992 and has remained around this level since then. Note: Percentage of GDP Source: The World Bank (2008) 139 2007 2005 2003 2001 Investment rate 1997 1995 1993 1991 1989 1987 1985 1983 1981 1979 1977 1975 1973 1971 1969 1967 1965 1963 Saving rate 1999 90 80 70 60 50 40 30 20 10 0 Saving and investment Figure 1. Saving and investment trends in Saudi Arabia 1963-2007 SEF 28,2 140 Before 1952, Saudi Arabia had a simple, but an inadequate financial system. This situation changed after the increase in oil prices in the 1970s-1980s when the government realized the need for an adequate and effective financial system. The system underwent major changes, including the establishment of the Saudi Arabian Monetary Agency in 1952, credit and other banking and financial institutions in the 1970s and the stock market in the 1980s. According to Albatel (2003), the recent reforms in the financial system in Saudi Arabia have impacted positively on the country’s economic growth and the financial reforms have affected the relationship between saving and investment. Thus, it becomes even more important to study the relationship between the two variables given these changes in the last few decades in Saudi Arabia. 3. The conceptual framework As indicated earlier, the correlation between saving and investment was first introduced as a criterion for assessing international capital mobility by Feldstein and Horioka (1980). The idea was that in a closed economy with low capital mobility, domestic saving finances all investment. However, in an open economy, domestic saving is not necessarily used to finance domestic investment, as saving will be used to gain better returns in international capital markets. In the words of Feldstein and Horioka (1980, p. 317) with perfect capital mobility, “there should be no relation between domestic saving and domestic investment: saving in each country responds to the worldwide opportunities for investment while investment in that country is financed by the worldwide pool of capital”. Feldstein and Horioka (1980) fit the following regression: IR ¼ a þ b SR ð1Þ where a is the constant, IR is the investment rate and SR is the saving rate. IR is defined as investment divided by GDP and SR is national saving divided by GDP. Their empirical finding for equation (1) using a sample of 16 countries from 1960-1974 yields a very high degree of correlation between the two rates, prompting the authors to conclude that capital is not very mobile among the major OECD countries. To examine the saving-investment correlation in Saudi Arabia as expressed in equation (1), we apply two cointegration techniques: ARDL bounds testing approach to cointegration and the Gregory and Hansen (1996) cointegration method with an endogenously determined structural break. The use of more than one estimator will ensure the robustness of the results. 4. Unit root tests Prior to conducting the cointegration tests, it is essential to check each time series for stationarity. If a time series is non-stationary, traditional regression analysis will produce spurious results. To ascertain the order of integration, we apply the traditional augmented Dickey-Fuller (ADF) and Phillips-Perron (PP) unit root tests. However, Perron (1989) shows that ignoring a structural break, as is the case with both ADF and PP tests can lead to the false acceptance of the unit root null hypothesis. Therefore, the Perron’s (1997) additive outlier (AO) model is applied, taking into account an endogenously determined structural break. When considering the AO model for testing a unit root, a two-step procedure is used. First the series is de-trended using the following regression: yt ¼ m þ bt þ gDT*t þ y~ t ð2Þ where y~ t is the de-trended series and DT*t ¼ 1ðt 2 T b Þ if t . Tb and zero otherwise. This procedure assumes that a structural break only affects the slope coefficient. Thus, the test is then performed using the t-statistic for a ¼ 1 in the regression: k X ci Dyt2i þ et ð3Þ y~ t ¼ ay~ t21 þ i¼1 These equations are estimated sequentially for all possible values of Tb (T b ¼ k þ 2; . . . ; T 2 1), where T is the total number of observations, so as to minimize the t-statistic for a ¼ 1. The null hypothesis is rejected if the t-statistic for a ¼ 1 is larger in absolute value than the corresponding critical value. The break date is assumed to be unknown and endogenously determined by the data. The lag length, k is data-determined using the general-to-specific method. 5. Cointegration tests The ARDL cointegration approach The cointegration concept is associated with the long-run equilibrium relationship between two or more variables. Commonly used methods for conducting cointegration tests include the residual-based Engle and Granger (1987) test and the maximum likelihood-based Johansen (1991, 1995) and Johansen and Juselius (1990) tests. Owing to low power and other problems associated with these tests, the ARDL bounds testing approach to cointegration has become popular in recent years[4]. The ARDL bounds testing procedure has numerous advantages over other cointegration techniques as outlined below: . The ARDL method avoids the pre-testing problems associated with the standard cointegration techniques. The pre-testing procedure in the unit root cointegration literature is problematic because the power of unit root tests are known to be typically low and there is a switch in the distribution function of the test statistics as one or more roots of the xt process approaches unity (Pesaran, 1997). . A further advantage of the ARDL method is that it is a more statistically significant approach for determining cointegrating relationships in small samples, whereas the Johansen cointegration technique requires larger samples to be valid (Ghatak and Siddiki, 2001). . By using the F-test, the ARDL cointegration method can distinguish which series is the dependent variable when cointegration exists (Narayan and Narayan, 2003). To investigate the existence of a long-run relationship in equation (1), we estimate the following unrestricted error-correction model (UECM): n n X X cj DI t2j þ d1 S þ d2 I t21 þ 11t bj DS t2j þ ð4Þ DS t ¼ a0 þ j¼1 DI t ¼ a0 þ n X j¼1 bj DI t2j þ j¼1 n X cj DS t2j þ f1 I t21 þ f2 S t21 þ 12t ð5Þ j¼1 where D is the first difference operator, S is the saving rate and I is the investment rate. The F-test is used for testing the existence of a long-run relationship between Saving and investment 141 SEF 28,2 142 these two variables through testing the significance of the lagged level of variables in the right-hand side of UECM. That is, we test for the null hypothesis of no cointegrating relation in equation (4), H0:d1 ¼ d2 ¼ 0 against the alternative HA:d1 ¼ d2 – 0. In equation (4), saving is the dependent variable and is expressed as F(S/I). In equation (5), where investment is the dependent variable, denoted as F(I/S), we test for the null hypothesis of no cointegration, H0:f1 ¼ f2 ¼ 0 against the alternative HA:f1 ¼ f2 – 0. These hypotheses are tested using the F-test with critical values tabulated by Pesaran et al. (2001). The asymptotic distributions of the F-statistics are non-standard under the null hypothesis of no cointegrating relationship between the examined variables, irrespective of whether the variables are purely I(0) or I(1), or mutually cointegrated. Pesaran et al. (2001) offer two sets of asymptotic critical values. The first set assumes that all variables are I(0) and the second set assumes that all variables are I(1). The null hypothesis of no cointegration is rejected if the calculated F-statistic is greater than the upper bound critical value. If the computed F-statistics is less than the lower bound critical value, then the null of no cointegration cannot be rejected. Pesaran et al. (2001) report the two sets of critical values based on 40,000 replications of a stochastic stimulation which provide critical value bounds for all classifications of the regressors into purely I(0), purely I(1) or mutually cointegrated for a sample size of 1,000 observations. However, Narayan (2005) computes the critical values for bounds for F-test for small samples sizes, which is also used in this study. Gregory-Hansen cointegration test The above ARDL bounds testing approach ignores the issue of any potential structural break in a cointegrating relationship. As with the unit root tests, Kunitomo (1996) argues that, in the presence of structural change, traditional cointegration tests which do not allow for a structural break may produce “spurious cointegration results”. Gregory and Hansen (1996) also show that the ADF test tends to “under-reject” the null hypothesis of no cointegration in the presence of a structural break. Gregory and Hansen address the problem of estimating cointegration relationships in the presence of a structural break by introducing a residual-based technique. The Gregory-Hansen methodology tests for the null hypothesis of no cointegration against the alternative of a cointegrating relationship in the presence of a potential break. The advantage of the Gregory and Hansen (1996) methodology is that, this testing procedure determines the time of the break endogenously[5]. In other words, the break point (TB) is unknown and is determined by finding the minimum values for the ADF t statistic. By taking into account the existence of a potential unknown and endogenously determined one-time break, the Gregory-Hansen approach allows for structural shifts in either the intercept alone, in both the trend and the level shift or for a full break. Thus, they present three models for testing cointegration allowing for the existence of a structural break in the cointegrating vector. The first model, known as a level shift model (Model C) contains an intercept and a level shift dummy as follows: y1t ¼ u1 þ u2 w1t þ a T y2t þ et t ¼ 1; . . . ; n: ð6Þ The second model (C/T) contains an intercept and a trend with a level shift dummy: y1t ¼ u1 þ u2 w1t þ bt þ a T y2t þ et t ¼ 1; . . . ; n: ð7Þ Saving and investment The third model is the full break model, called a regime shift (C/S); this allows for changes in both intercept and slope, as follows: y1t ¼ u1 þ u2 w1t þ aT1 y2t þ aT2 y2t wtt þ et t ¼ 1; . . . ; n: ð8Þ 143 Model C/S includes two dummy variables, one for the intercept and one for the slope. In the context of our analysis, y1t and y2t are the saving and investment rates; u1 and a1 are the intercept and slope coefficients before the shift; u2 and a2 denote the changes to the intercept and slope coefficients at the time of the shift. The dummy variable is denoted by w1r and is defined by: w1t ¼ 0; if t # ½ht
and w1t ¼ 1; if t . ½ht
where the unknown parameter t 1(0,1) denotes the relative timing of the change point. 6. Empirical findings Unit root results This study uses annual data from 1963 to 2007 for Saudi Arabia. The data for gross domestic saving, gross domestic investment and GDP are obtained from the World Bank (2008), World Bank World Tables. The data for gross domestic saving and gross domestic investment are measured as a proportion of GDP, consistent with the Feldstein and Horioka (1980) study. To ascertain the order of integration, we first conduct the traditional ADF and PP unit root tests. These tests suggest that both the variables in the model are non-stationary (Refer to Table I). Since the ADF and PP tests suffer from power deficiency in the presence of a structural break, we also apply Perron’s (1997) AO model to the data. The results reported in Table II also find evidence of unit root (non-stationary), confirming the results computed by the ADF and PP tests. The estimated break date of 1998 for Description Saving rate Investment rate ADF test statistics Result Unit root Unit root 21.3592 22.5947 PP test statistics 2 1.4261 2 2.6991 Result Unit root Unit root Note: Critical value at the 5 percent level is 2 3.5155 Description k Tb Saving rate Investment rate 8 6 1998 1979 Test statistic 2 2.6774 2 3.5523 Table I. ADF and PP unit root test results Result Unit root Unit root Notes: Critical values at the 1, 5 and 10 percent are 2 5.45, 24.83 and 24.48, respectively; the maximum k ¼ 8 Table II. AO model for determining the break date SEF 28,2 144 saving and 1979 for investment correspond with real events in Saudi Arabia. The 1979 break for the investment rate coincides with the oil price shock, the Iranian revolution and the energy crisis. The structural break for the saving rate in 1998 coincides with the economic and financial reforms that took place in Saudi Arabia. During this period, the country experienced large budget deficits associated with low oil prices worldwide. The local currency was also subject to a wave of speculation leading the government to intervene to prop up the currency (Dibooglu and Aleisa, 2004). This period also coincided with the Asian financial crisis of 1997 which negatively affected Saudi Arabia in particular and the Gulf region in general. Cointegration results To investigate the existence of a long-run relationship in equation (1), we estimate equations (4) and (5). As explained earlier, the ARDL bounds testing procedure involves the comparison of the computed F-statistics with the critical values. Both the computed F-statistics of 1.7538 and 4.1662 (given in Table III) are less than the lower bound critical values given by Pesaran et al. (2001) and Narayan (2005). As the computed F-statistics are below the lower bound critical value at the 5 percent significance level, the null of no cointegration cannot be rejected. Therefore, no evidence exists for a long-run relationship between saving and investment in Saudi Arabia. The Gregory-Hansen procedure for cointegration in the three models (equations 6-8) is estimated to test for the existence of a long-run relationship between saving and investment with an endogenously determined structural break. The results and the critical values are reported below in Table IV. The results for all the three models (C, C/T, and C/S) indicate that the null of no cointegration cannot be rejected at the 5 percent significance level. The break dates of 1976, 1983, and 1984 detected by the Gregory-Hansen procedure correspond with the oil crisis leading to the world recession, and the reforms that occurred in Saudi Arabia during the mid 1970s and early 1980s[6]. This finding is consistent with the result of the ARDL bounds testing cointegration approach, leading to conclusion that no long-run relationship exists between saving and investment. As per Feldstein and Horioka, evidence of no correlation between the two Table III. F-test for testing the long-run relationship between saving and investment Computed F-statistics (FBounds) 2 F(S/I) Computed F-statistics (FBounds) 2 F(I/S) Critical bounds for n ¼ 45 from Narayan (2005) Critical bounds from Pesaran et al. (2001) UCB: 7.91 UCB: 7.30 Note: Critical bounds from the two authors are from Table CI v Case V with unrestricted intercept and trend in the model Model Table IV. Gregory-Hansen cointegration test with structural break 1.7538 4.1662 LCB: 7.08 LCB: 6.56 C C/T C/S Break point ADF Critical value at 5% 1976 1983 1984 2 2.62 2 4.11 2 2.92 24.61 24.99 24.95 Note: The null hypothesis being no cointegration between saving and investment Source: Critical values are provided by Gregory and Hansen (1996) Result Do not reject H0 Do not reject H0 Do not reject H0 variables indicates a high degree of international capital mobility for Saudi Arabia. This result is consistent with many other studies including AbuAl-Foul (2006), Chaudhri and Wilson (2000), Sachsida and Mendonça (2006), and Kollias et al. (2008) but it contradicts the studies by Jansen (1996), Isaksson (2001), and Ang (2007). Overall, our empirical results indicate that no relationship exists between saving and investment in the long-run and thus, capital is highly mobile in Saudi Arabia. This is plausible given the country has undergone tremendous financial and economic reforms during the last few decades These reforms have led to massive capital inflows, dominated by oil revenue flows to the Saudi government. However, other investment and private inflows into the assets markets have also increased recently, especially with the country’s accession to the World Trade Organization in 2005. Further to this, foreign direct investment sharply increased registering at US$18 billion in 2006, with the stock of foreign direct investment accounting for US$48 billion in 2006 (Al-Jasser and Banafe, 2009). 7. Summary and conclusion The aim of the paper is to examine the long-run relationship between saving and investment as a criterion for assessing international capital mobility. Saudi Arabia is a suitable case given that, it is the largest economy in the Middle Eastern and Arab nations and a heavily oil-dependent country. The approach adopted here follows the Feldstein and Horioka (1980) study and applies various stationarity and cointegration tests. These include the traditional ADF and PP unit root tests as well as the Perron’s (1997) stationary test with an endogenously determined structural break. For robustness, we also apply two cointegration procedures, to test for the long-run relationship between the saving and investment, the ARDL bounds testing procedure and the Gregory and Hansen (1996) cointegration method. Our empirical results from both the cointegration procedures suggest that no long-run relationship exists between saving and investment, indicating the presence of a high degree of capital mobility in Saudi Arabia for the 1963-2007 period. The absence of this relationship is plausible as the country has undergone financial and economic reforms leading to massive capital inflows, which are dominated by oil revenue flows to the Saudi government. This chain of events has been reinforced with the country’s accession to the World Trade Organization during 2005 which saw increases in other investment and private inflows into assets markets. The effect of capital being highly mobile for an oil dependent country of Saudi Arabia is consistent with the Feldstein and Horioka proposition. But, this conclusion cannot be generalized for all oil-dependent countries and thus further research on the capital mobility issue is needed for other oil-dependent countries. Notes 1. Austria, Belgium, Germany, Greece, Italy, Luxembourg, Spain, and the UK. 2. Egypt, Jordan, Morocco, and Tunisia. 3. Broad money is defined as M3 (which consists of money circulation plus deposits savings and other deposits) plus quasi-monetary deposits. Saving and investment 145 SEF 28,2 4. Numerous cointegration studies employ the ARDL model instead of the traditional maximum likelihood test based on Johansen (1988) and Johansen and Juselius (1990). These studies include Bahmani-Oskooee and Nasir (2004), Narayan (2005), and Kollias et al. (2008). 5. The Gregory and Hansen (1996) test is applicable only for I(1) processes. The ADF, the PP and the AO unit root tests all show that both the saving rate and investment rate are I(1). 146 6. The break dates given by the Perron (1997) unit root test are somewhat different from those given by the Gregory and Hansen (1996) cointegration test. The different break dates occur because the Perron (1997) unit root test searches for a break in individual series, while the cointegration test searches for a break in the residual of two series. References AbuAl-Foul, B. (2006), “The relationship between savings and investment in MENA countries”, paper presented at the 42nd Annual MBAA/ABE Conference in Chicago, IL. Albatel, A. (2003), “Money, finance and economic growth the case of Saudi Arabia”, available at: www.erf.org.eg/cms.php?id¼conferences_details&conference_id¼5 (accessed October 24, 2008). Al-Jasser, M. and Banafe, A. (2009), “Capital flows and financial assets in emerging markets: determinants, consequences and challenges for central banks”, available at: www.bis.org/ publ/bppdf/bispap44v.pdf (accessed February 12, 2009). Al-Rajhi, A., Al-Salamah, A., Malik, M. and Wilson, R. (2003), Economic Development in Saudi Arabia, Routledge, London. Ang, J. (2007), “Are savings and investment cointegrated? The case of Malaysia (1965-2003)”, Applied Economics, Vol. 39, pp. 2167-74. Apergis, N. and Tsoulfidis, L. (1997), “The relationship between saving and finance: theory and evidence from E.U. countries”, Research in Economics, Vol. 51 No. 4, pp. 333-58. Arginon, I. and Roldan, J. (1994), “Saving, investment and international capital mobility in EC countries”, European Economic Review, Vol. 38 No. 1, pp. 59-67. Bahmani-Oskooee, M. and Nasir, A. (2004), “ARDL approach to test the productivity bias hypothesis”, Review of Development Economics, Vol. 8 No. 3, pp. 483-8. Baxter, M. and Crucini, M. (1993), “Explaining savings investment correlations”, American Economic Review, Vol. 83 No. 3, pp. 416-36. Chaudhri, D. and Wilson, E. (2000), “Savings, investment, productivity and economic growth of Australia 1861-1990: some explorations”, The Economic Record, Vol. 76, pp. 55-73. Coakley, J., Kulasi, F. and Smith, R. (1996), “Current account solvency and the Feldstein-Horioka puzzle”, Economic Journal, Vol. 106 No. 436, pp. 620-7. Cooray, A. and Sinha, D. (2007), “The Feldstein-Horioka model re-visited for African countries”, Applied Economics, Vol. 39 No. 12, pp. 1501-10. Dibooglu, S. and Aleisa, E. (2004), “Oil prices, terms of trade shocks, and macroeconomic fluctuations in Saudi Arabia”, Contemporary Economic Policy, Vol. 22 No. 1, pp. 50-62. EIU (2007), Country Profile Saudi Arabia, Annual Report, Economist Intelligence Unit, London. Engle, R. and Granger, C. (1987), “Co-integration and error correction: representation estimation and testing”, Econometrica, Vol. 55 No. 2, pp. 251-76. Feldstein, M. and Horioka, C. (1980), “Domestic savings and international capital flows”, The Economic Journal, Vol. 90, pp. 314-29. Ghatak, S. and Siddiki, J. (2001), “The use of ARDL approach in estimating virtual exchange rates in India”, Journal of Applied Statistics, Vol. 28 No. 5, pp. 273-583. Gregory, A. and Hansen, B. (1996), “Tests for cointegration in models with regime and trend shifts”, Oxford Bulletin of Economics and Statistics, Vol. 58 No. 3, pp. 555-60. Ho, T. (2000), “Regime switching investment-savings correlations and international capital mobility”, Applied Economics Letters, Vol. 7 No. 9, pp. 619-22. Isaksson, A. (2001), “Financial liberalisation, foreign aid, and capital mobility: evidence from 90 developing countries”, Journal of International Financial Markets, Institutions and Money, Vol. 11 No. 3, pp. 309-38. Jansen, W. (1996), “Estimating saving-investment correlations: evidence for OECD countries based on an error correction model”, Journal of International Money and Finance, Vol. 15 No. 5, pp. 749-81. Johansen, S. (1988), “Statistical analysis of cointegration vectors”, Journal of Economic Dynamic and Control, Vol. 12, pp. 231-54. Johansen, S. (1991), “Estimations and hypothesis testing of cointegration vectors in Gaussian vector autoregressive models”, Econometrica, Vol. 56 No. 6, pp. 1551-80. Johansen, S. (1995), Likelihood-Based Inference in Cointegrated Vector Autoregressive Models, Oxford University Press, New York, NY. Johansen, S. and Juselius, K. (1990), “Maximum likelihood estimation and inference on cointegration with application to the demand for money”, Oxford Bulletin of Economics and Statistics, Vol. 52 No. 2, pp. 169-210. Kasuga, H. (2004), “Savings-investment correlations in developing countries”, Economics Letters, Vol. 83 No. 3, pp. 371-6. Kollias, C., Mylonidis, N. and Paleologou, S. (2008), “The Feldstein-Horioka puzzle across EU members: evidence from the ARDL bounds approach and panel data”, International Review of Economics and Finance, Vol. 17, pp. 380-7. Kunitomo, N. (1996), “Tests of unit roots and cointegration hypotheses in econometric models”, Japanese Economic Review, Vol. 47 No. 1, pp. 79-109. Miller, S. (1988), “Are saving and investment cointegrated?”, Economic Letters, Vol. 27 No. 1, pp. 31-4. Murphy, R. (1986), “Productivity shocks, non-traded goods and optimal capital accumulation”, European Economic Review, Vol. 30 No. 5, pp. 1081-95. Narayan, P. (2005), “The savings and investment nexus for China: evidence from cointegration tests”, Applied Economics, Vol. 37 No. 1, pp. 1979-90. Narayan, P. and Narayan, S. (2003), “Savings behaviour in Fiji: an empirical assessment using the ARDL approach to cointegration”, Discussion Papers 02/03, Monash University, Melbourne. Perron, P. (1989), “The great crash, the oil price shock, and the unit root hypothesis”, Econometrica, Vol. 57 No. 16, pp. 1361-401. Perron, P. (1997), “Further evidence on breaking trend functions in macroeconomic variables”, Journal of Econometrics, Vol. 80 No. 2, pp. 355-85. Pesaran, M. (1997), “The role of economic theory in modelling the long-run”, Economic Journal, Vol. 107 No. 1, pp. 178-91. Pesaran, M., Shin, Y. and Smith, R. (2001), “Bounds testing approaches to the analysis of level relationships”, Journal of Applied Econometrics, Vol. 16 No. 3, pp. 289-326. Sachsida, A. and Mendonça, M. (2006), “Domestic saving and investment revised: can the Feldstein-Horioka equation be used for policy analysis?”, Discussion Papers 1158, Instituto de Pesquisa Econômica Aplicada – IPEA. Saving and investment 147 SEF 28,2 148 SAMBA (2008), “Saudi’s Arabia’s 2008 budget, 2007 performance”, available at: http://www. samba.com/GblDocs/sa2008_en.pdf (accessed November 28, 2008). Wong, D. (1990), “What do savings-investment relationships tell us about capital mobility?”, Journal of International Money and Finance, Vol. 9 No. 1, pp. 60-74. (The) World Bank (2008), World Bank World Tables, The World Bank, Washington, DC. About the authors Reetu Verma is a Lecturer in the School of Economics at the Faculty of Commerce at the University of Wollongong. She completed her Honours, Master’s and PhD in Economics at the University of Wollongong, Australia. She has published extensively in reputable international journals (for example, ASEAN Economic Bulletin, The Middle East Business & Economic Review, and South Asia Economic Journal ). Ali Salman Saleh is a Senior Lecturer and Coordinator for postgraduate studies by research at the School of Economics and Finance, Victoria University, Australia. He completed his Master’s and PhD in Economics at the University of Wollongong, Australia. Dr Saleh has published extensively in reputable international journals (for example, Journal of Policy Modeling, Singapore Economic Review, and the Asia Pacific Journal of Economics and Business). His current research interests concentrate on the areas of applied economics, and development issues in small and medium enterprises. To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints