1. Introduction
The GCC (Gulf Cooperation Council) countries relied heavily on hydrocarbon in the last couple of decades to grow rapidly and maintain better living standards. From 2000 to 2017, hydrocarbon contributed 80 percent of the government revenue, 65 percent of the total exports and 42 percent of the GDP of the GCC countries [
1]. The over-reliance on hydrocarbon swelled public investment and discouraged private investment in the GCC countries. Although rich hydrocarbon resources generated sufficient revenue for the government in the GCC countries to avoid economic meltdown in the short run, but the dwindling hydrocarbon reserves, and severe oil price fluctuation put the GCC countries in a difficult position [
2]. Scholars suggest that the GCC countries need urgent reforms and more space for private investment in order to maintain economic stability. In the first leg the GCC countries opened their doors for foreign investment while in the second leg they heavily invested abroad in order to connect local firms to global chain of production and thus diversify their economies away from hydrocarbon.
There is a widespread literature on the impact of incoming foreign direct investment inflows FDI in the GCC countries [
3,
4,
5]; however, the impact of outbound investment on the GCC countries economic well-being is still a mystery, especially when it comes to discover links between OFDI and domestic investment in the GCC countries. Empirical literature suggest that outbound investment can potentially increase local investment given that investing abroad link local investors to the global chain of production [
6,
7,
8,
9].OFDI can positively affect private investment in source countries if (1) OFDI is financed from the abundant saving and bulging foreign exchange reserves of a country and (2) OFDI increase the return of local firms by employing the most efficient factors abroad [
10]. In cases where OFDI is motivated to reduce cost, then the outcome can be different. High cost and lack of opportunities encourage firms to relocate production abroad at the cost of local investment. In other words, the two forces of ‘complement’ and ‘substitute’ make it difficult to generalize the impact of OFDI on domestic investment [
7,
11,
12,
13].
Therefore, in this study, our main objective is to understand whether the recent surge of OFDI increase or decrease private investment in the GCC countries keeping in view that private investment is the main cog in the process of economic diversification both in developed and developing countries? Following [
8] and [
14], we assume that OFDI complement private investment and diversify economic activities in the GCC countries. In case OFDI decreases private investment, then OFDI can degrade the financial structure and the efforts of economic diversification in the GCC countries [
10]. Private investment is crucial for economic progress due to its flexibility, connectivity and its absorptive capacity of latest ideas and information. Private investment also plays the role of a buffer against the economic shocks and uncertainties (IMF, 2018). For example, [
15] suggested that private investment is more productive and innovative than any other form of investment. The findings of this study provide policy direction to the GGC countries on how to better utilize their scared financial resources.
The rest of the study is organized as follows.
Section 1 discusses OFDI, FDI and Capital formation in the GCC countries while
Section 2 presents literature review.
Section 3 explains the data sources and econometric methodologies.
Section 4 reports the results and findings of the study and
Section 5 concludes it.
1.1. Research Hypothesis
Our main research question and hypothesis to understand the impact of OFDI on private investment in the GCC countries can be described as:
Hypothesis 1 (H 1): OFDI positively contribute to the private investment in the GCC countries;
Hypothesis 2 (H 2): OFDI adversely affect private investment in the GCC countries.
1.2. OFDI, FDI and Domestic Capital Formation in GCC Countries
The GCC is a regional intergovernmental political and economic union consisting of Bahrain, Kuwait, Saudi Arabia, Qatar, Oman and the UAE. The charter of the GCC was signed in 1981. Over the course of almost four decades, the GCC has gradually evolved into an integrated economic bloc. Hydrocarbon is the prima facia and the backbone of the GCC bloc. The government revenue, total exports and GDP growth squarely counts on hydrocarbon proceeds.
Hydrocarbon price increased many folds since the charter of the GCC was signed but it never stayed stable. Hydrocarbon prices exhibited severe volatility, e.g., oil price, the main mix of hydrocarbon, increased from
$ 80 per barrel in 2010 to
$ 106 per barrel in 2014. By 2016, the oil prices were hovering around
$ 30 per barrel; it suddenly jumped to
$ 70 per barrel by 2018. High volatility in the prices of hydrocarbon feeds economic instability and uncertainty in the GCC countries [
16]. This problem amplify when we include the changing patterns of global production and the discovery of new energy sources in the equation and consider that the attraction of hydrocarbon is on decline. In this backdrop, the GCC countries need diversification to revamp their economies [
17].
According to [
18], outbound foreign investment in the GCC countries can indeed be the beginning of a successful transition from hydrocarbon to a diverse economy. Recently, the GCC countries widely invested abroad in strategic sectors such as computations, hotel chains and financial firms as well as in transport and communication, etc. Moreover, one of the basic motives of these endeavors is to promote domestic investment at home.
Table 1 shows that OFDI in the GCC countries increased over time in absolute terms; however, the share of OFDI as percent of GDP did not. Rather OFDI as a percent of GDP registered decline from 3.8 percent in 2008 to 2.1 percent of the GDP in 2017. Yet, this decline in OFDI in the GCC countries is less than the decline in the incoming FDI.
The shares of public and private investment in the GCC countries are generally pitched against each other and normally the dominant control of the government is blamed for sluggish private investment in the GCC countries [
1,
17]. Governments in GCC countries hold a major chunk of the local economy and therefore it is considered a hurdle in the process of diversification [
1]. However,
Table 1 shows that public and private investment in GCC countries as a percent of GDP did not change a lot in the last decade. From 2008 to 2017, private investment in the GCC countries remained within the bounds of 32 to 34 percent of the GDP while public investment vacillated in the range of 20 to 23 percent of the GDP in the same period of time.
Table 1 also shows that oil and gas revenue as a percent of GDP exhibit great volatility. Therefore, relying on hydrocarbon is risky business.
2. Literature Review
The empirical literature on the nexus between OFDI and capital formation in home countries is broadly divided into two genres such as complementary and substitute [
9].OFDI complements private investment if it link local firms with the global chain of production. Similarly, OFDI can promote domestic investment if it increase the return of the local firm by connecting them to efficient labor and capital abroad [
8]. However, when OFDI relocate production from home to host countries then it adversely affect domestic investment [
19,
20,
21,
22,
23].
The studies supporting the positive and complementary role of OFDI in promoting domestic investment assumes that the main motives of OFDI is efficiency seeking. Efficiency seeking OFDI link the home and the host countries economic activities in such a way that they exploit the economies of scale, reduce costs of production and increase efficiency. Efficiency seeking vertical OFDI complements trade by relocating parts of production chain from home to the host country [
24]. Vertical OFDI boost trade in intermediate products between host and home countries and help the home country to access latest technology and technical knowhow [
25]. Vertical OFDI do not disrupt domestic production and thus play the role of conduit to promote investment in source countries [
26,
27,
28].
On the other hand, a number of studies looking at the adverse (substitutions) impact of OFDI suggest that contrary to the impact of vertical OFDI, horizontal OFDI retard production in home country by relocating domestic production abroad. Such OFDI (horizontal) is supposed to substitute home country production with the host country production and thus adversely affecting local investment. Some studies are more critical of OFDI and suggest that OFDI not only relocate domestic production abroad, but it also reduce economic activities at home by replacing domestic production with imports [
26] while other explored that OFDI squeeze scarce financial resources at home country by venturing out at the cost of local investment [
8,
10,
23,
29].
There are four motives for firms seeking to invest abroad such as efficiency-seeking, resource seeking, market-seeking and strategic asset-seeking. The strategic asset–seeking motive of OFDI is of significant importance. Strategic asset seeking OFDI is supposed to improve the domestic investment in source and host countries by bringing in new knowledge and technologies in high spillover sectors [
30,
31,
32,
33]. A number of studies underscored these theoretical underpinnings. Regarding the aggregate OFDI, there is a positive long-run unidirectional causal relationship running from OFDI to domestic investment in China [
14]. Similarly, there is strong evidence of positive spillover impact of OFDI on private investment not only on country level, but also on inter-regional level in China [
34]. On the contrary, the macroeconomic cross country analysis was conducted, and results show that increase in OFDI decrease domestic investment [
22].Same findings are confirmed by United States, Japan, Germany and the United Kingdom [
35]. A similar analysis was conducted for the United States and Germany where they discovered positive long-run effect of OFDI on domestic investment in the US, however, their finding in case of Germany was mixed. They found that outward FDI has positive short run and negative long-run effect on domestic private investment there. This means that OFDI in Germany contributed to domestic private investment in the short run and substitute domestic private investment in the long run [
29].
Literature based on neoclassical and Keynesian outlooks view public and the private investment differently [
36,
37]. Neoclassical support private investment and assume that private capital is the main force which increase efficiency, stimulate creativity and promote diversity. Neoclassical disregard public capital in promoting economic growth and in some cases even they blame public investment for inefficiency [
15]. From here a perception emerges which assume that public and private investment generally crowd–out each other and thus clubbing them together can potentially increase the problem of aggregation bias. However, the Keynesians refute this argument and they suggest that public investment actually pave the way for effective utilization of private investment and thus public and private investment crowd in each other. In this study, we want to see whether public investment crowd–in or crowd–out private investment in the GCC countries.
3. Methodology and Data
The selection of an appropriate model for empirical analysis in panel data has huge impact on the final outcome. A correct model produces an efficient and consistent results while an incorrect model lead to wrong conclusion. For example, until recently, most of the cross country studies in economics literature wrongly assumed that errors are independently distributed across data. While in reality variables in cross country studies count on each other, particularly in the long run. Therefore, to drop the erroneous assumption of independent errors in cross country panel data and get consistent results, many studies moved away from traditional techniques like panel dynamic OLS [
38], panel fully modified OLS approach [
39] and panel pooled and mean group tests [
40] to new techniques such as cross sectional-ARDL [
41,
42], GMM [
43], FGLS [
41,
44] and PCSE tests [
45].These models not only address the issue of cross country dependence, but they also solve the problems of heteroscedasticity and serial correlation in the panel data. For example, the use of inappropriate techniques report no impact of foreign remittances inflows on labor productivity in [
46], however, as soon they addressed the issue of cross country dependence and used an appropriate model, they found a strong and significant impact of foreign remittances on labor productivity.
Our methodological approach is to pool cross-sectional time series. The time series dimension of the pooled data is more problematic than cross-section because observations in timer series are usually not independent [
47]. OLS normally indicate an ideal situation where the errors are independent and homoscedastic, but in reality, the error terms deviates from such assumptions and therefore need a thorough analysis in application of an appropriate model. The GCC countries are closely connected both in policy orientation, culture and economic activities. They depend on each other in many ways and therefore the possibility of cross-sectional dependence in the GCC countries, especially in case of FDI inflows and outflows, cannot be ruled out. [
48] and [
49] consider that Cross-correlation occurs very frequently due to spatial spillover effects, omitted common factor and inter-actions within socioeconomic network. We consider that outbound foreign direct investment (OFDI) in one country is not independent of the outbound foreign direct investment in another country in the GCC countries. Same is true for FDI inflows and therefore we suspect the problem of cross-sectional dependence in our panel series.
Panel-based studies normally use ARDL or GMM approached. Although the panel GMM techniques address the issue of endogeneity, but it does not work efficiently in the presence of cross-sectional dependency and structural breaks. System GMM also has limitations in studies based on small N (cross section) and large T (time dimension). On the contrary, Panel CS–ARDL not only take care of endogeneity problem, but it also addresses the issue of cross sectional dependence and capture the long and short run impacts of the variables. However, CS–ARDL technique has certain limitations. It is considered that CS–ARDL is valid only for a panel data with long cross sectional and short time series (N>T) [
50].
Knowing that our time series (25 years) is longer than our cross sectional (6 countries), we dropped the idea of using CS–ARDL and looked for alternate techniques. Literature suggest a number of approaches to address the issue of cross-sectional dependence when T>N in panel series. Among them the two, i.e., feasible generalized least squares (FGLS) [
41,
44] and panel-corrected standard errors (PCSE) [
45] received wide attention. Therefore, in this study we rely on FGLS and PCSE techniques (
Appendix A).
Though FGLS model is widely popular among the researchers, but [
45] identify two important issues in the FGLS model. First, the FGLS by default can only be estimated when the number of time periods (T) is more than the cross section (N). Otherwise, the associated EVCM (error-variance covariance matrix), the base of the FGLS models, cannot be inverted with small T and large N. Second, the FGLS model tends to produce unacceptably small standard error estimates which render hypothesis testing useless [
45,
50]. At the same time with cross-section heteroskedasticity the OLS standard errors are also inconsistent. Therefore, to address these issues, panel-corrected standard errors (PCSE) convincingly demonstrate that PCSE large-T asymptotic-based standard errors, which correct for contemporaneous correlation between the subjects, perform well in small panels [
45]. PCSE preserves the (Prais–Winsten) weighting of observations for autocorrelation but uses a sandwich estimator to incorporate cross-sectional dependence when calculating standard errors. In order to control of heterogeneity, serial correlation and cross sectional dependence in our panel data-based models [
50], we regress the following baseline model using the techniques of PCSE (FGLS is used for robustness check).
where, i stands for cross-sectional dimension; i = 1……i and time period t = 1…….t and αi represents country specific effects. εit is a random disturbance term of mean zero. αi stands for country specific effects. It is pertinent to mention that the disturbance term in equation (1) is consist of two components, i.e., time invariant and idiosyncratic error components. We assume that the time invariant component is specific to the individual and does not change over time while the idiosyncratic error component is usually assumed to be independent of both the regressors and the individual error components.
DPI in equation (1) represents the domestic private investment in the GCC countries while OFDI is outbound foreign direct investment originating from the GCC countries. Z is a control variable for GDP growth, interest rate (IR), incoming FDI (IFDI), trade openness (TO), public investment (PUB) and incoming FDI, etc. later we will expand the set of our control variables by introducing the variable of public investment (PUB). Public investment is considered one of the biggest hurdles for private investment in the GCC countries. Therefore, the inclusion of public investment will not only better reflect the economic condition of the GCC countries, but it will also contribute to the robustness check of the key results. The final version of our model is given as:
We avoid using lagged dependent variable (LDV) on the right-hand side of the equation in this study because the lagged dependent variables in panel estimations other than Generalized Method of Moments (GMM) produce bias coefficient [
51].
Table 2 presents the whole list of our dependent and independent variables along with their source and theoretical justification. The data are available on demand. The range of our data is from 1993 to 2017.
4. Result and Discussions
Descriptive statistics give us a clear and generalized view of the data set. Therefore, we start our discussion with reporting descriptive statistics in
Table 3 where the average level of OFDI as a percentage of GDP is 1.45 while the average level of IFDI as a percentage of GDP is 2.50. This shows that foreign investment in the GCC countries on average is higher than the GCC countries investment abroad. Normally, it is perceived that strong government control means more public and less private investment, but average public investment in the GCC countries as a percent of GDP is almost half that of average private investment (
Table 3). The GCC countries registered 5.25 percent growth rate on average in the last two and half decades. OFDI, IFDI, IR and other variables report relatively small variation across the panels; however, there is big difference in the trade as share of GDP across the GCC countries.
The GCC countries are closely linked and their economic interdependence has further increased in the last few decades. The peculiar nature of the GCC countries gives way to suspension that our selected variables may suffer from the problem of cross-sectional dependence. Therefore, we use [
56] CD (cross sectional dependence) test to investigate contemporaneous correlation in our panel series. The null hypothesis in the [
56] CD test is cross-sectional independence while the alternative hypothesis suggest presence of cross-sectional dependence in the panel data. CD test values in
Table 4,
Table 5 and
Table 6 reports highly significant cross-sectional dependence in the data across the board.
Keeping in view the diverse nature and size of the GCC countries, we applied the Breusch–Pagan test of homoscedasticity in order to determine whether our data sets exhibit constant or non-constant variance in cross-sections. The Breusch–Pagan test statistic in
Table 4 for the baseline and extended models are highly significant and thus they reject the null hypothesis of homogeneity in data. The data also report a serial correlation problem. This means that the cross-sectional dependence, heteroscedasticity and serial correlation problem in our dataset leave the traditional panel estimation techniques redundant. The only option left are feasibly generalized least square and panel corrected standard errors (PCSE) test. FGLS and PCSE produce efficient and consistent estimators in the case of cross-sectional dependence and group hetero in the data set. Though our main purpose of this study is to explore the impact of OFDI on private investment, but we start our analysis with the impact of OFDI on overall investment in the GCC countries. The main reason of doing this is to better understand the contribution of OFDI to private investment in the GCC countries.
The FGLS and PCSE results in
Table 4 suggest that the impact of OFDI, GDP growth and trade openness on overall investment in the GCC countries are insignificant. In
Table 4 only incoming FDI assert a positive and significant influence on the overall investment. The insignificant impact of GDP growth rate and trade on overall investment in
Table 4 indicates that the GCC countries relied on natural resources in their export to rest of the world while the products required at home were imported from abroad. Therefore, despite many-fold increase in trade and GDP, the domestic investment is still out of touch with the two most important macroeconomic variables. Interest rate (IR) is another important variable which carry a correct sign but assert insignificant influence on domestic investment in the GCC countries.
In light of the above findings now we come to our main question. Does OFDI affect private investment in GCC countries?
Table 5 report the impact of OFDI, IFDI and other variables (in model 2) on domestic private investment in the GCC countries. The results show that contrary to OFDI impact on the overall investment, the impact of OFDI on private investment in the GCC countries is positive where one percent increases in OFDI increases private capital from 0.23 percent of 0.31 percent. This shows that OFDI complement private investment. This finding, when compared with the impact of OFDI on overall investment, provides a policy direction that OFDI can be instrumental in diversification of the GCC countries.
Table 5 also shows that GDP growth did not influence private capital formation and this finding is in line with [
15] who shows that one-way causality runs from private investment to GDP while GDP does not cause private investment. Trade impact on private investment is positive and significant, but not very strong. On average, a percent increase in trade enhances private investment by 0.03 percent, which is marginal. This shows that the trade in the GCC countries is not properly connected to global chain of production. Yet, when we look into the big chunk of hydrocarbon in the GCC export and the efforts of the GCC countries to substitute imports with local products, the results seems satisfactory. The impact of IFDI in
Table 5 is significant, but only under PSCE models. Simply,
Table 5 shows that OFDI can be an effective tool to promote private investment in the GCC countries.
The positive and significant role of public sector in private investment in
Table 5 indicates that public sector crowd in private investment in the GCC countries. This finding negates the crowd–out hypothesis of public private investment of neoclassical and supports the crowd–in hypothesis of the Keynesian which shed positive light on the role of government in providing the private investment with the required infrastructure and the necessary support. The presence of public sector also enhances the values of OFDI impact on private investment and it reaches maximum to 0.4 in the presence of public capital (in model 3 of
Table 5).
From
Table 4 and 5 we can draw a conclusion that OFDI does not affect overall investment in the GCC countries, but OFDI strongly affect private investment in the GCC countries. OFDI impact on private investment in
Table 5 is higher than IFDI impact on private investment and therefore, OFDI can be used as a policy instrument to diversify local economies in the GCC countries. Now to take the discussion a bit further, we replace private investment (DPI) by public investment (PUB) in Equation (2). The main purpose of this exercise is to understand whether OFDI and private investment has any impact on public investment.
Table 6 reports that in the base line model (under FGLS and PCSE) OFDI does not affect public investment, only IFDI and GDP growth and interest rate significantly affect public investment. However, the impact of the GDP growth and IFDI is quite different. GDP is adversely affecting public investment while IFDI is positively contributing to it. This shows that as GDP of the GCC countries grow, the share of public investment decrease.
Table 6 shows that interest rate does not affect public investment in the GCC countries.
In the extended models, only private investment under FGLS model significantly and positively affects public investment which again shows that public and private investment are mutually crowding in each other in the GCC countries. The mutual positive impact of the public and private investment on each other not only confirms the crowding-in Keynesian hypothesis, but it is also a good reflection of the reality in the GCC countries where public sector investment in infrastructure asserted a positive spill over impact on the private economy [
1]. Public sector investment in the GCC countries increases the confidence of private investors, reduce their cost of doing business and it provides them assurance that the government support is behind them.