Relationship between foreign capital flows, domestic investment and savings in the SADC region

ABSTRACT

The importance of foreign capital in the domestic economy cannot be underestimated as it bridges the gap between domestic capital demand and supply. Given this background the paper studies the relationship between the different types of foreign capital flows in the Southern Africa Development Region (SADC) region – foreign direct investment (FDI), remittances, cross border bank flows (CBF), overseas development assistance (ODA) – and domestic savings and investment, employing the panel cointegration test and the dynamic ordinary least squares method (DOLS). The empirical results reveal that there is a strong positive relationship between domestic investment and domestic savings, FDI and remittances. These findings indicate that FDI remittances help in overcoming the limits on the domestic capital formation in the SADC region through permitting a rate of investment which is in excess of that which can be generated by domestic savings. Important policy implications on attracting foreign capital flows are discussed in the paper.

KEYWORDS:

• Foreign capital flows
• domestic investment
• domestic savings
• panel cointegration
• Southern Africa Development Community

JEL CLASSIFICATION:

• F21
• F30
• P33

1. Introduction

Rising levels of foreign capital in the form of foreign direct investment (FDI), foreign portfolio investment (FPI), cross border bank flows, overseas development assistance (ODA) and, recently, remittances have become an issue of great concern to the development debate in developing countries. Ndikumana & Verick (2007) and Macias & Massa (2009) highlight that there is evidence indicating that there has been a rise in the different types of capital flows in Africa since the 1990s. The authors highlight that FDI inflows grew by 15% between 2004 and 2007. The same applies to portfolio equity and bond flows which grew up by more than 340% over the same period. At the same time, cross border bank flow activity grew significantly (Macias & Massa, 2009).

Even though there has been a rise in capital flows to African countries, Ndikumana & Verick (2007) highlight that their impact on investment, economic growth and alleviation of poverty is still an issue of great concern. In the case of the Southern Africa Development Community (SADC) region, the International Council on Social Welfare (2015) indicates that as of 2015 approximately half of the population survived on less than $1 per day. This is despite the region receiving an increase in FDI inflow of 40 times from US$372 million in 1980 to 17 billion in 2008 (Wentworth et al., 2015). It is also interesting to note that the SADC (2012) report shows that investment in the region has been fluctuating. In 2005, it stood at 20.3; in 2006, it was 19.6 and increased to 23.5 in 2008, but decreasing again in 2010 to 22.2. As of 2012, the rate had dropped to 20.9%.

However, it is important to take into account that the role of foreign capital flows on the domestic economy is ambiguous. The available literature on the subject presents two views. The first set of studies such as Boon (2000); Athukorala & Sen (2002); Waheed (2004); Holmes (2005); Caceres & Saca (2006); Osili (2007); Zhu et al. (2009); Das & Serieux (2010); Balde (2011); Hossain (2014) suggests that foreign capital flows can play a very important role in developing countries as a source of development finance. These studies argue that foreign capital may act as a vehicle which can stimulate economic growth in developing countries. Much of these studies have relied on the two-gap model which points to gaps between savings and investment needs, with foreign capital flows bridging the gaps, resulting in higher levels of investment and, ultimately, an increase in economic growth. According to these studies foreign capital has the capacity to compliment domestic savings, promoting investment and hence higher levels of economic growth in the receiving countries.

On the other hand, despite the positive effects associated with foreign capital flows as highlighted above, studies such as the Economic Commission for Africa (ECA, 2006) report and Prasad et al. (2007) argue that external financial flows may have negative effects on African economies through the ‘Dutch Disease’ effects in which the increase in the flow of capital increases the demand for the domestic currency which will, in turn, reduce the competitiveness of a country's export industry and make imports cheaper, thereby deteriorating the country's external position. In addition, external financial flows may increase the vulnerability of a country which is caused by the high volatility and unpredictability of capital flows. Furthermore, there are other studies (Waheed, 2004; Morton et al., 2010; Balde, 2011; Hossain, 2014) which have argued that foreign capital weakens the link between domestic investment and savings, a model which has come to be known as the Feldstein–Horioka hypothesis.

Given the conflicting views in the literature, the paper thus seeks to provide empirical evidence that sheds light on the role of the different types of foreign capital flows on domestic investment and savings in the SADC region. A distinguishing feature of this study is the inclusion of four different types of external financial flows which is contrary to the available studies which have mostly focused on FDI only.

To empirically investigate the impact of the different forms of capital flows on domestic savings, data from the 15 SADC countries is used for the period from 1980 to 2013. The study employed panel cointegration, specifically the Kao and Pedroni tests for robust results and the dynamic ordinary least squares model (DOLS) to analyse the long term relationship between the variables of interest. The results indicate that of the four different types of foreign capital flows, FDI, cross border bank flows and remittances are important determinants of investment in the SADC region. The results also indicate that there is a bi-directional causality between domestic investment and savings. This suggests that foreign capital flows permit the SADC countries to overcome the limits in the domestic economy, allowing the countries to realise a level of investment which is higher than can be generated by domestic savings.

2. Theoretical framework and empirical literature review

The study augments Pagano (1993), Bailliu (2000) and Agosin & Machado’s (2005) models. The Agosin and Machado model assumes that foreign capital enters into the existing basket of gross capital formation in the hosting economy:(1) Equation (1) shows that investment I at time t is comprised of domestic investment () and foreign investment ().

Domestic investment is regarded as a stock variable, which is defined as the difference between desired and actual capital stock. The model is thus specified as:(2) Where is the desired capital stock by domestic investors, while is a stock adjustment coefficient, with .

The desired amount of capital is also regarded as a function of the expected growth () as well as the difference (y) between actual output (Y) and full capacity output ():(3) where

Assuming that the capital stock depreciates at a rate of () per period, gross investment equals:(4) Incorporating the role of savings in the investment process, Bailliu (2000) points out that financial intermediaries are responsible for transforming savings into investment. In doing this, they absorb resources so that a dollar saved by households will generate less than a dollar's worth of investment. It is also assumed that a fraction, , of each dollar saved is available for investment, whereas is retained by the financial intermediaries as a reward for the services supplied. In the closed-economy version of the model, capital market equilibrium requires that the fraction of savings by domestic residents left after financial intermediaries have taken their share must equal gross investment. Thus, equilibrium in the domestic capital market ensures that:(5)

Extending equation (5) to take into account the effect of foreign capital on investment, given that foreign investors invest through financial intermediaries resulting in a larger pool of savings available for investment, in the presence of international capital flows the capital market equilibrium becomes:(6) where S is domestic savings, NCF_t represents net international capital flows and I is domestic investment. Bailliu (2000) highlights two channels through which foreign capital may promote investment in the domestic economy. Firstly, capital flows can promote domestic investment in the domestic economy if it is not used for consumption. In addition, investment financed by foreign capital must not crowd out domestically financed investment. Secondly, capital flows can foster investment growth in the domestic economy if they lead to investments that are associated with positive spillovers. These positive externalities include competition which will result in domestic firms becoming more productive. It is also believed that foreign firms can help restructure key sectors of the economy by making them more competitive and export oriented (United Nations Centre of Transnational Corporations [UNCTC] 1995).

Despite the much applauded positive impact of foreign capital flows on investment, Griffin (1970) points out that foreign capital in the form of aid, as much as it can be argued that it is tied to a specific project and it is not possible to switch aid from investment to consumption, is ‘fallacious’. The author argues that when the country is given aid for a project, it is recognised that foreign capital donated would not be financing the project to which it is apparently tied to, but the marginal investment project. Foreign capital, therefore, finances not the marginal investment project, but the marginal expenditure project and usually, expenditure on the margin, according to Griffin (1970), is likely to be a consumption good. Thus according to this model, foreign capital in the form of aid promotes consumption at the expense of investment.

The empirical literature on the nexus between foreign capital, savings and investment in Africa is sparse. Among the few available studies, the methodology adopted raises a number of questions especially how they deal with the occurrence of endogeneity in the data set. Some studies focused on theoretical discussions while others applied empirical analysis using various types of statistical tests. The available studies in this regard, mainly from Asia and Western countries, include Chen (1977), Rana & Dawiling (1990), Boon (2000), Athukorala & Sen (2002), Waheed (2004), Holmes (2005), Caceres & Saca (2006), Osili (2007), Zhu et al. (2009), Das & Serieux (2010), Balde (2011), Hossain (2014). These studies illustrate that foreign capital leads to an increase in investment through the savings channel. However, some of these studies suggest that foreign capital displaces domestic savings, promoting consumption and hence a reduction in investment.

Of the available studies, Employing the Common Correlated Effects mean group (CCEMG) model, Hossain (2014) examined the differential impacts of foreign capital and remittance inflows on domestic savings of 63 countries for the period 1971 to 2010. From the conventional homogeneous panel estimates, foreign aid and remittances were found to be substituting domestic savings. However, when the author utilised the CCEMG model, which considers cross-section dependence and parameter heterogeneity properties, the empirical results suggest that foreign aid and remittance flows had a significant negative impact on domestic savings, with the latter having a stronger crowding-out effect on domestic savings compared to the former. The impact of FDI and portfolio inflows was found to be statistically insignificant. According to the author, the results of the study supported the hypothesis that large capital inflows lead to displacement of domestic savings. Thus assuming that savings are a major determinant of investment, an incremental contribution of foreign capital is likely to impact negatively on investment in the domestic economy. These findings are in line with Kumar (2007).

Of the majority of studies which examined the impact of remittances on savings, Athukorala & Sen (2002), in the case of India established that remittances displace domestic savings, even though the impact was found to be weak. This result was found to be consistent with Caceres & Saca (2006) for El Salvador who found that for the 1990s, increased remittance inflows into the country were accompanied by a sharp decline in savings in that country. On the contrary, Osili (2007) established that remittances have a positive impact on savings, investment and economic growth through reducing poverty in the country of origin.

There are other studies which have suggested that foreign capital flows promote domestic savings and hence investment. Of these studies, Ndikumana & Verick (2007) examined the effects of FDI on domestic investment in the Sub-Saharan Africa and established that FDI crowds in domestic investment. The author highlighted that countries are likely to gain more through implementing measures which are aimed at improving the domestic investment climate.

In another study, Amadou (2011) on a study on Togo established that FDI and loans are the main channels through which foreign capital can positively influence domestic investment. However, with regards to foreign portfolio investment, the authors found that its coefficient is negative even though insignificant. These results imply that the effect of foreign capital on investment largely depends on its type. These results are consistent with Boswarth & Collins (1999).

In the case of the SADC countries, there are fewer studies which have been carried out to examine the link between foreign capital and investment. The majority of the studies have rather examined the effect of the different types of foreign capital on economic growth or savings (Kapingura et al., 2015; Kapingura et al., 2016; Mugowo, 2017). The available studies are country specific, mostly South Africa for example Temitope (2014). The study thus contributes by analysing the effect of the different forms of foreign capital on domestic investment in the SADC region.

3. Data and empirical methodology

The data utilised in the study is for the period between 1980 and 2013. The data utilised in the study was obtained from the United Nations Conference on Trade and Development, World Bank African Development Indicators and Bank of International Settlements (BIS). The period of study was chosen based on data availability. The empirical model utilised in the study is similar to the work of Balde (2011) which is as follows:(7) where I_it is investment of country i at time t. The X variable contains the host of other investment determinants variables to be employed in the study. Taking into account all other important variables utilised in the study, the empirical model is estimated as follows:(8) where I is Investment, FEF is external financial flows, OPEN is openness, GDP is gross domestic product, GDI is gross domestic savings, INF is inflation, i is interest rate, FD is financial development, and ε is the error term.

3.1. Definition of variables and a priori expectations

The FEF represents foreign capital flow. It is measured by foreign direct investment (green investments), overseas development assistance (economic infrastructure), remittances and cross border bank flows. However, as in Kapingura et al. (2015), there is no consensus regarding the impact of foreign capital flows on investment as indicated in the literature. One strand of literature suggests that foreign capital flows help to overcome the limits on the domestic capital formation and permits a rate of investment in excess of the national savings. Thus in this case, a positive relationship is expected between investment and foreign capital flow. A positive relationship is also expected between investment and openness. A more open economy is likely to attract more investment. GDP represents economic growth. A positive relationship between GDP and investment is expected. This emanates from the theoretical argument that an increase in income increases the propensity to save with the savings channelled towards investment (Mohamed, 2003). Thus a positive relationship is also expected between gross savings and investment. The INF is equal to the average growth rate in the consumer price index and captures stability of the macroeconomic environment. A negative relationship between inflation and investment is expected. The I is the rate of interest, based on the Keynesian model, a negative relationship between investment and interest rate is expected. An increase in interest rate increases the costs of borrowing. Financial development (FD) is measured by bank credit to the private sector (DCP) as a ratio of GDP and money supply as measured by M3 as a percentage of GDP. A positive relationship between investment and financial sector development is expected. There are two strands of theory which explain the impact of financial development on investment as in Kapingura et al. (2015). The authors highlight that financial sector development enhances resource mobilisation, increasing savings and hence investment, assuming that all resources saved are channelled towards investment. On the other hand, as pointed out by Nair (2006), financial sector development may boost consumption given that it makes access to finance easier.

3.2. Testing for stationarity in panel data

In order to determine the order of integration of the variables utilised, the study employed the Levin & Lin (LL) test and the Im Pesaran & Shin (IPS) test. The two tests were utilised to obtain robust results. De Wet & Van Eyden (2005) point out that the IPS method has better small sample properties and is more intuitive in its construction as compared to the LL test. In addition, the IPS test does not require values which are Monte Carlo simulation based.

3.3. Panel cointegration test

The study utilised the Kao (1999) cointegration test and the Pedroni residual based panel cointegration test to examine if there is a long-term relationship between the variables of interest. These two tests are discussed briefly.

3.3.1. The Kao panel cointegration test

Morshed (2010) highlights that Kao (1999) presents two tests to examine the presence of cointegration. The two tests are based on the Dickey–Fuller and Augmented Dickey–Fuller tests. As in Morshed (2010), considering a bivariate model of the following form:where is the fixed effect which varies across the cross-section observations, is the slope parameter, and are independent random walks for all , the residual series should be I(1) series (Morshed, 2010). The Dickey–Fuller and the Augmented Dickey–Fuller (ADF) tests can be estimated on the residuals, respectively, as:

However, Kapingura et al. (2015) highlights that Kao's panel cointegration test, though applauded on its ability to impose homogenous cointegrating vectors and AR coefficients, has been criticised for its failure not to allow for multiple exogenous variables in the cointegrating vector.

3.3.2. The Pedroni panel cointegration test

Morshed (2010) highlights that Pedroni (1999) proposes seven panel cointegration test statistics of which four are based on the within-dimension and the other three are based on the between-dimension. Kapingura et al. (2015) points out that Pedroni's tests have been applauded for allowing multiple (m = 1, 2, … , M) regressors, for the cointegration vector to vary across different sections of the panel, and for heterogeneity in the errors across cross-section units.

Morshed (2010) shows that the null hypothesis for the within-dimension statistics of no cointegration can be stated as:

On the other hand, for the between-dimension statistics, the null hypothesis of no cointegration can be stated as:

Morshed (2010) further highlights that in the first stage we compute the regression residuals from the hypothesised cointegration regression, which can be of the following form:where T refers to the number of observations over time, N refers to the number of the individual members in the panel, and M refers to the number of regression variables. In this case x and y are assumed to be integrated of order one. The slope coefficients and the slope varies across the individual cross-sections. For a full discussion of the seven Pedroni statistics see Pedroni (1999:658).

3.3.3. Dynamic ordinary least squares method

Having established the presence of cointegration in the model, the next step was to analyse the long-term relationship between the variables of interest. The study employed the dynamic ordinary least squares method (DOLS) in the analysis.

Hawdon & Ahmed (1997) argue that the DOLS approach improves on OLS by coping with small samples and dynamic sources of bias. The authors also argue that the method is a robust single equation approach which corrects for regressors’ endogeneity by the inclusion of leads and lags of first differences of the regressors, and for serially correlated errors by a GLS procedure. In addition, the method has the same asymptotic optimality properties as the Johansen distribution.

In addition, Akmal & Stern (2001) also argue that DOLS, as opposed to many other estimators, does not require all the individuals’ series in a long-term relationship to be integrated of order one, as it is also applicable to systems involving variables of differing orders of integration. In the case of I(1) series, the DOLS technique involves regressing one variable on the contemporaneous levels of the other variables and on the leads and lags of their first differences and a constant term. Given that the individual variables of the model are all I(1), the DOLS estimating equation is written as:(12) Thus having established the presence of cointegration, equation (12) was estimated to establish the long-term relationships between the variables of interest.

4. Presentation of empirical results

The time series properties of the panel data were initially examined through the LL and the IPS tests and the results are reported in Table 1.

Table 1. Panel unit root tests: level series.

Panel	GFCF	GDS	GDP	I	INF	DCP	M2	FDI	CBF	REM	OPE
LL	0.8573	−1.7143	−1.2853	−0.3136	−0.7572	1.49033	1.0283	−0.7868	2.7009	1.2513	−0.5005
IPS	1.8754	−1.9303	−0.7302	−0.4866	−0.5127	2.30491	1.3185	−0.9540	−0.4215	0.8761	−2.4030
MW	0.2547	−2.2931	−0.5390	−1.4557	−0.9907	2.5572	2.1987	−0.9983	0.6683	−0.7546	0.92879

Notes: GFCF = gross fixed capital formation; GDS = gross domestic savings; GDP = gross domestic savings; I = interest rate; INF = inflation; DCP = domestic credit to the private sector; M2 = Monetary aggregate 2; FDI = foreign direct investment; CBF = cross border finance; REM = remittances; OPE = openness. LL, Levin & Lin test; IPS, Im, Pesaran & Shin test; MW, Maddala and Wu.

(1) The t-statistics for each test are reported.

(2) The null hypothesis: variable has a unit root.

Source: Author's calculations

As reported in Table 1, all the variables employed in the model were found to be non-stationary. This suggests that the variables have a unit root. The variables were further analysed as first difference and the results are reported in Table 2.

Table 2. Panel unit root test: first difference series.

Panel	GFCF	GDS	GDP	I	INF	DCP	M2	FDI	CBB	REM	OPE
LL	−6.585***	−7.581***	−11.41***	−9.353***	−10.29***	−5.002***	−6.930***	−15.08***	−4.080***	−5.4157***	−7.788***
IPS	−7.874***	−10.77***	−16.39***	−11.68***	−15.58***	−7.539***	−7.799***	−15.74***	−7.636***	−4.5124***	−10.08***
MW	−9.584***	−15.861***	−16.47***	−14.38***	−17.76***	−12.55***	−11.04***	−17.81***	−12.26***	−7.4872***	−3.110***

Notes: GFCF = gross fixed capital formation; GDS = gross domestic savings; GDP = gross domestic savings; I = interest rate; INF = inflation; DCP = domestic credit to the private sector; M2 = Monetary aggregate 2; FDI = foreign direct investment; CBF = cross border finance; REM = remittances; OPE = openness. LL, Levin & Lin test; IPS, Im, Pesaran & Shin test; MW, Maddala and Wu.

(1) The t-statistics for each test is reported.

(2) The null hypothesis: variable has a unit root.

(3) The null hypothesis of stationarity is tested. *, **, *** indicate the parameters that are significant at 10%, 5%, 1% probability, respectively.

Source: Author's calculations

Table 2 indicates that all the variables become stationary after first differencing. This is a requirement for both the Kao and the Pedroni tests that the variables should be integrated of order 1. Thus, having established the order of integration of the series, the next step is to examine if there is cointegration between the series as per equation (8) based on the Kao and Pedroni models. The results of the two different models are reported in Tables 3 and 4, respectively. Based on the Kao cointegration test there is evidence of cointegration between investment as measured by the gross fixed capital formation (GFCF) in the SADC region and its determinants since the ADF test is significant at 5% level as shown in Table 3.

Table 3. Kao cointegration test results of the savings function.

	t-statistic	P-value
ADF	−2.72806**	0.043
Residual variance	0.00780
HAC variance	0.00083

Notes: ADF, Augmented Dickey–Fuller; HAC, Heteroskedasticity and Autocorrelation Consistent covariance matrix estimator.

(1) Null hypothesis: No cointegration.

(2) The lags used were automatically selected based on the Newey–West bandwidth using the Bartlett kernel.

(3) The null hypothesis of no cointegration is tested. ** indicates the parameters are significant at 5% probability.

Source: Author's calculations.

Table 4. Pedroni cointegration test results on the savings function.

Tests (Pedroni, 1999)	Standardised statistics	P-values
Panel v statistic	−0.734059	0.7685
Panel p statistic	−0.677620	0.2490
Panel pp statistic	−1.973654	0.0242**
Panel adf statistic	−3.227043	0.0006***
Group p statistic	0.601690	0.7263
Group pp statistic	−2.154682	0.0156***
Group adf statistic	−3.175499	0.0007***

Notes: (1) Panel v is non-parametric variance ratio statistic. Panel p and panel pp are analogous to the non-parametric Phillips–Perron p and t-statistic, respectively. Panel adf is a parametric statistic based on the ADF statistic. Group p is analogous to the Phillips–Perron p statistic. Group pp and group adf are analogous to the Phillips–Perron t-statistic and the ADF statistic, respectively.

(2) The formulae for calculating these statistics can be found in Pedroni (1999), Table A1.

(3) The null hypothesis of no cointegration is tested. *, **, *** indicate the parameters that are significant at 10%, 5%, 1% probability, respectively.

Source: Author's calculations.

This confirms that there is a long-term relationship between the variables of interest. Influencing any of these variables is likely to influence the level of investment in the SADC region. This result is also confirmed by the Pedroni test, as shown in Table 4, which indicates that almost all of the statistics are significant, confirming again that there exists a cointegrating relationship between investment and the selected variables in the model.

Having established cointegration, the DOLS model was utilised to estimate the long-term relationships between the variables of interest (see Table 5 for results).

Table 5. Dynamic ordinary least squares (DOLS) investment function results for SADC countries, 1980–2013.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
FDI	0.0281***		0.0286***	0.0248***	0.0300***
CBF	0.0042*	0.0063*		0.0087*	0.0055*
REM	0.0708**	0.0705**	0.0718**		0.099***
ODA	0.0791	0.0798	0.0790	0.0889
OPE	0.0207*	0.0238*	0.0204*	0.0198*	0.0262*
GDP	0.0424**	0.0399	0.0427**	0.0433**	0.0479*
GDS	0.0476**	0.0538	0.0488**	0.0144**	0.0507**
INF	−0.003**	−0.0029**	−0.0031**	−0.0029*	−0.0028*
I	−0.0147*	−0.0140**	−0.0148**	−0.0145**	−0.0138**
M2	0.027**	0.033**	0.029**	0.025**	0.012**
DCP	0.051**	0.041**	0.040**	0.032**	0.035**
R^2	0.75	0.64	0.63	0.56	0.57
Adjusted R^2	0.72	0.60	0.61	0.50	0.53
Observations	442	442	442	442	442
Jarque–Bera test: 1.2756(p value: 0.5785)

Notes: FDI = foreign direct investment; CBF = cross border finance; REM = remittances; ODA, overseas development assistance; OPE = openness; GDP = gross domestic savings; GDS = gross domestic savings; INF = inflation; I = interest rate; M2 = Monetary aggregate 2; DCP = domestic credit to the private sector.

(1) The DOLS was estimated including two lags and one lead.

(2) All regressions include a constant not reported.

(3) *, **, *** denote significance at the 10%, 5% and 1% level, respectively.

Source: Author’s calculations.

The long-term results indicate that the three types of external financial flows – FDI, CBF and remittances –contribute positively to investment in the SADC region. On the other hand, ODA is found to be insignificant though positive. These results suggest that FDI, CBF and REM contribute positively to the generation of physical capital in the SADC region. In other words, these three types of external finance complement domestic savings. These results are consistent with the findings of Mohamed (2003) who established that foreign capital in the form of FDI supplements investment. Thus in this case, foreign capital helps to overcome the limits on the domestic capital formation as it permits a rate of investment which is in excess of that which can be generated by domestic savings. Following FDI, which is significant at the 1% level, remittances emerge as another important source of development finance in the SADC region through supporting investment. This is consistent with Balde (2011) who suggests that remittances may have indirect positive effects on growth and poverty through investment. The author points out that the investment can be done by households or migrants themselves.

On the other hand, the negative relationship between ODA and investment in the SADC region is likely due to the bilateral aid which many of the countries in the region receive and it is usually tied politically and economically. Cassen et al. (1986) suggest that aid tying is likely to raise the cost of investment, especially where the tied aid biases the recipient investment towards excessive capital intensity. This is also supported by Griffin (1970) who highlighted the fungibility of aid or the switching of foreign savings into uses which are not productive. Griffin also points out that aid may finance not the high-priority projects but rather the more marginal ones.

Of the other control variables, the results suggest that openness (OPE) is consistent with the a priori expectation and carries the correct sign. This suggests that more open economies are likely to realise more investments. This is consistent with Balde (2011). The empirical results also suggest that savings are another important determinant of investment in the region. These results are consistent with Balde (2011) and Kapingura et al. (2015). This also supports the Harrod–Domar model which emphasises the positive role of savings towards investment and hence economic growth. The empirical results also suggest that inflation is negative and highly significant, suggesting that an inflationary environment brings uncertainties. Investors will not be willing to take their money where there is uncertainty. This recommends therefore that stable and low inflation rates are another important factor determining the rate of investment in the region. The interest rate carries the appropriate expected sign and is consistent with previous studies such as those by Mohamed (2003) and Balde (2011). This also corroborates the Keynesian model which emphasises low interest rates as a way of promoting investment.

The empirical results also reveal that both measures of financial development are positively related to investment in the long-term. This is consistent with Love & Zicchino (2006), Almeida & Campello (2007), Misati & Nyamongo (2011) and Huang (2011). These studies point out that a well-developed financial system provides more opportunities and incentives to invest. In addition, well developed financial systems are characterised with low fund costs, a large variety of instruments and low agency costs. The process of diverting savings to investment is more effective in the case of developed financial systems. To further analyse the relationship between savings, investment and foreign financial flows, panel Granger causality tests were performed as represented in Table 6.

Table 6. Panel Granger causality tests based on the panel vector auto regression (PVAR) Model in first difference (sample period: 1980–2013).

Dependent variable	Independent variables
	ΔInvestment	ΔSavings
ΔFDI	4.1426 (0.0301)**	5.8249 (0.0415)**
ΔCBF	0.0916 (0.0552)*	0.4996 (0.0790)*
ΔODA	6.7432 (0.0469)	5.7862 (0.0324)
ΔREM	5.9429 (0.0512)**	4.3494 (0.0136)**
ΔGDS	5.5339 (0.0629)*
ΔGFCF		3.5138 (0.0425)**

Notes: FDI = foreign direct investment; CBF = cross border finance; ODA, overseas development assistance; REM = remittances; GDS = gross domestic savings; GFCF = gross fixed capital formation.

(1) The optimum lag for each system was selected based on the Schwarz information criterion.

(2) *, ** denote rejection of H₀ at 10% and 5% level of significance, respectively.

(3) The F-statistic is reported with the significance level (). The H₀: No causality.

The Granger causality test results show that there is a short-term bi-directional causal effect running from FDI, CBF and REM to investment and vice versa. The ODA is insignificant. The savings variable is significant though at a weaker level (10%). The weaker level of the short-term causality from savings to investment implies a high degree of short-term international capital mobility in the SADC region. This suggests that much of the investment in the region was contributed partially by foreign financial flows. This, therefore, proposes that the inflow of FDI, CBF and remittances are of great importance in the region. The results also show that the three types of external financial flows (FDI, CBF and remittances) also contribute towards savings. The results are in sharp contrast to the Feldstein–Horioka prediction of foreign capital weakening the link between investment and savings. Investment is also significant and contributes towards domestic savings. This result is consistent with Boon (2000) who suggests that an increase in investment enhances aggregate output and national income; this will in turn induce further savings.

5. Summary and conclusion of the study

The study focused on examining the nexus between investment, savings and foreign capital flows in the SADC region considering that the region is experiencing low levels of investment as well as low levels of savings. The study employed panel cointegration tests and the dynamic ordinary least squares model for the long-term model. The empirical results revealed that FDI, CBF and remittances featured as important determinants of investment in the region. However, ODA was found to be insignificant towards investment. Granger causality tests were further carried out to analyse the causal relationship between investment, foreign capital flows and savings in the SADC region. The results highlighted that there is evidence of bi-directional causality between savings and investment in the SADC region.

The results highlighted that investment in the region apart from savings is also financed by foreign capital flows. Thus foreign capital helps in overcoming the limits on the domestic capital formation as it permits a rate of investment which is in excess of that which can be generated by domestic savings. The findings also suggest that as much as aid is important in the SADC region, when it is tied politically and economically it may not be productive. Financial development also featured as another important factor determining investment in the SADC region. The empirical results recommend that countries in the region should endeavour to achieve macroeconomic stability with the aim of stimulating the inflow of foreign capital. In addition, there is a need to further develop the financial sector given its positive role on investment as well as harnessing remittances which can further encourage people to save and invest.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Forget Mingiri Kapingura http://orcid.org/0000-0002-5808-5612