Foreign Direct Divestment Phenomenon in Selected Sub-Saharan African Countries

Abstract

Foreign direct divestment can occur for either external or internal factors. The determinants of FDI are also the same determinants for FDD. FDD might lead to numerous negative economic factors such as a decline in economic development, reduction in employment and might also cripple the facilitation in technology transfers. In this paper, the FDD concept in the Sub-Saharan African countries was investigated using annual data spanning from 1998 to 2018. The panel autoregressive distributive lag was used to develop the FDD model. The findings of the panel ARDL long run equation revealed that lending rates and urbanisation have a negative and significant influence on foreign direct investment. Further, the findings revealed an insignificant influence of real gross domestic product per capita on FDI. Finally, trade openness showed a positive significant impact on foreign direct investment. We recommend policies that increase FDI through the cost of borrowing since increasing this results in foreign direct divestment. Real gross domestic product per capita cannot be used for policy making purposes in the study. Trade openness makes a country more accessible on the world market and thus, policies that promote foreign trade such as exporting complex and sophisticated products, trade liberalisation, free trade agreements and open trade systems could help reduce the presence of foreign direct divestment in the selected countries. Finally, urbanisation deter foreign direct investment, therefore countries should invest more on infrastructure and reduce poverty in rural areas to transform them into urban areas to decrease urbanisation.

1. Introduction

Foreign investment and foreign trade remain interesting subject matters for most researchers, economists and governments owing to their prestigious impact on a country’s aggregate economy and its sectors. With that being mentioned, notably less attention has been given to foreign direct divestment (FDD). FDD seems to be the most neglected area of research and it seems as if there is not much literature regarding this phenomenon. This study aims to address the FDD concept in selected Sub-Saharan African countries. Foreign direct divestment is a concept involving an adjustment in the ownership of a business that involves the partial or full disposal of an asset or a business unit [1]. Also, when there is an increase in the general prices of goods, it is referred to as inflation and the opposite as deflation. The same concept also applies when there is an increase in investment inflows from foreign nationals to a domestic economy, it is referred to as FDI inflows for the host economy and the opposite is referred to as FDD. This implies that foreign direct divestment occurs when there is a decrease in foreign direct investment. Most researchers, policy makers and governments seem to be mostly concerned about the trends in FDI while totally neglecting putting measures that may reduce or sidestep FDD to have consistent and stable FDI inflows and outflows.

FDD can occur for either external or internal factors. García-Bolívar [1] suggests that weak business climate seemingly contributes to a decision to divest as much as there is no proven direct correlation. FDD might lead to numerous negative economic factors such as a decline in economic development, reduction in employment and might also cripple the facilitation in technology transfers. Among other reasons, this might be because most developing and Sub-Saharan African countries use FDI to fill the gap between domestic investment and savings due to their low levels income. According to literature, the determinants of foreign direct investment are the same for foreign direct divestment but with the opposite sign. However, no consensus seems to have been reached regarding the determinants of FDI according to past studies [2].

Boddewyn found that foreign direct divestment is the opposite of FDI. Before divestment, there must be investment and there are also several studies across the globe on foreign direct investment, such as studies by Kumari and Sharma [2]; Tahmad and Adow [3].

Most studies have already established that there is investment but tend to pay little or no attention to the concept of divestment. This paper therefore aims to fill research gaps in literature by attempting to find any occurrences of foreign direct divestment during the period 1998–2018. The limited availability of data restricted the study from including all developing and all Sub-Saharan African countries. The countries under investigation are Botswana, Egypt, India, Namibia, Nigeria and South Africa. This study attempted to find variables that are most likely to cause foreign direct divestment. The chosen variables are real gross domestic product, trade openness, lending rates and urbanisation. The rest of the paper is therefore organised as follows, following the introductory section is Section 2 which presents the trends of FDI in the selected Sub-Saharan, developing and emerging countries. Section 3 presents the theoretical analysis of the macroeconomics of FDI and FDD. Section 4 gives a brief review of literature followed by Section 5 which discusses the methodology of the study. Section 6 focuses on the discussion of the empirical results and the last section concludes the study and provides recommendations.

2. FDI trends in the selected Sub-Saharan, developing and emerging economies

Figure 1 shows the trends of FDI inflows as a percentage of GDP for six countries from 1998 to 2018. The trends give an idea whether these countries are failing to attract FDI or it is merely investors losing interest. In the year 1999, Namibia was experiencing better inflows reaching a peak of roughly 11% than all other countries until it rapidly dropped in the year 2001. In South Africa, FDI inflows increased rapidly from 0.71% in 2000 to 6% and then significantly dropped within a year and this occurrence of foreign direct divestment was highly perceptible.

Figure 1.

FDI inflows as a percentage of GDP for the 1998 to 2018 period [4].

Botswana’s vulnerability to external economic shocks was experienced during the global financial crisis of 2008. FDI inflows declined as signposted by [5] due to the low global demand for minerals, which led to a sharp decline in commodity prices and volumes traded and this translated into provisional closures of diamond mines, and possible job losses. The country’s dependence on minerals which are finite left it more susceptible to external economic shocks and this further entails that the country is nothing without its diamond mining. Also, evidence pointed out that Botswana has gone from being low-income nation to an upper-middle income country after the discovery of the diamond mine, with an increased real GDP averaged at 4.6% between 1994 and 2011 [5].

All countries experienced a foreign direct divestment after the world financial crisis in 2008 of which Namibia and Egypt mostly recovered in 2010 and 2011 respectively. This was the most severe financial crisis the world economy has ever faced since the Great Depression. These countries need to find ways that attract foreign investment to earn foreign currency and bring employment opportunities to their respective economies. In addition, Namibia’s inflows seem to be upward trending over the years moving away from the mean average, say 1.5%. However, FDI inflows to developing countries remained quite stable since 2004 with an increase of roughly two percent relative to a decrease of 27 percent for developed economies [6].

The world recession owing to the global financial crisis was highly perceptible in 2008 as shown in Figure 2. This occurrence of foreign direct divestment was not only experienced by developing and emerging economies as discussed earlier in Figure 1 but the world economy as well. The world economy experienced another fall in FDI inflows from 2016 and this decline seem to be noticeable again in 2018 onward. In addition, world FDI flows continuously declined by 13 percent in 2018. This decline was due to United States (US) intercontinental enterprises through returns at US of tax reforms and foreign earnings [7].

Figure 2.

World FDI inflows (US$ trillion) for the 1998 to 2018 period [4].

The occurrence of foreign direct divestment in 2018 could also have been induced by the spillover effects of the US-China trade wars which could have affected the world economy through trade openness (measured by the sum of imports and exports) and made it difficult for some countries (particularly developing countries) to be accessible in the world market. US has since imposed tariffs of around $436 billion on Chinese goods since July 2008 until July 2019 and this trade war was significant and most likely to effect in some economic displacements [8]. The spillover effect of the US-China trade war is that both these countries affect any significant trade, economic and investment links and any direct foreign investment links with their respective trading partners [8].

3. The macroeconomics of foreign direct investment and foreign direct divestment: theoretical analyisis

The FDI trends discussed earlier in Section 2 have precisely shown the movement of foreign direct investment changing its direction heading to foreign direct divestment. This was due to the global financial crisis and the spillover effects of the US-China trade wars. That said, Maček [9] postulated that the dynamic economic progresses in the flow of economic and financial crisis have shown the likelihood of foreign direct investment changing its course resulting to FDD. This implies that FDI and FDD cannot occur simultaneously in a single economy, but each may occur in either South Africa, Nigeria, Egypt etc. The determinants of FDD are said to reciprocate those of FDI but with the opposite sign as suggested by Boddewyn [10]. This theory was developed from Dunning’s theory of FDI. This implies that theoretical analysis of FDI concept also applies to FDD but in reverse. There are various studies that paid too much attention on FDI discussions but ignore the other side of the coin. This is like a detective who focuses on how to catch criminals but does not investigate the reasons for crime and how to prevent it. Here is what theory says concerning the rationale for both foreign direct investment and foreign direct divestment:

3.1 Dunning’s FDI theory

There are three basic conditions that must be met for FDI to take place as postulated by Dunning [11]. (i) a firm has net competitive exclusive advantages in relation to other countries in serving specific markets. That said, these exclusive advantages are at least for a period notably to the firm or country owning them and mostly take form of the ownership of tangible assets. (ii) Assuming that the first condition is met, the firm owning these advantages must benefit when utilising them on its own relative to leasing or selling them to foreign firms. This implies that it must benefit through internalisation from its own activities in rather than externalising them through contracts and or licencing to foreign firms. (iii) If condition (i) and (ii) are assumed to have been satisfied, the firm must be profitable when utilising these advantages in alignment with a minimum of one factor input (natural resources included) outside its geographical borders. If not, Dunning stated that domestic markets would be served solely by domestic production and foreign markets by exports.

These three conditions explain the rationale for a firm to engage in FDI and international production activities if it has more ownership advantages, the better the incentive it holds to internalise them and the more it finds it more profitable to exploit them beyond its geographical boundaries [11]. This implies that these conditions are interrelated and must be satisfied simultaneously, and if not, FDI activities are more likely to change direction to divestment.

3.2 Boddewyn’s FDD theory

Boddewyn’s [10] used Dunning’s theory of FDI to develop the foreign direct divestment concept. Boddewyn stated that violating at least one of Dunning’s conditions of FDI would lead to foreign direct divestment. FDD occurs when: (i) a firm no longer has net competitive advantages over firms of other countries; (ii) or if it has the net competitive advantages, they are no longer beneficial for self-use rather than renting or selling them to foreign firms; (iii) or the firm no longer realise profits from using its internalised net competitive advantages beyond its national boundaries (i.e., the firm now finds it profitable to produce locally to domestic markets and export the surplus to foreign markets). This theory is quite simple and straightforward since it was adopted from Dunning’s FDI theory. If one pays close attention, the differences between these theories is the opposite sign and the keywords “and” and “or” for the FDI and FDD theories, respectively. As mentioned earlier, Dunning’s theory requires all the conditions to be met, hence the word “and” while Boddewyn’s theory requires at least one condition [10]. Also, violating at least one of the Dunning’s conditions to FDI leads to FDD. Therefore, these two theories extensively explain the rationale for both FDD and FDI to take place.

4. Literature review

This section is divided into theoretical (which adds to the theories already discussed in Section 3) and empirical literature reviewed by the study.

4.1 Theoretical review

4.1.1 The eclectic paradigm theory

Based on Dunning’s three conditions in the Eclectic Paradigm, Boddewyn [10] found that foreign direct divestment occurs when a firm no longer enjoys the net competitive advantages over firms of other economies; or when it has competitive advantages but is not profitable to adopt these advantages; or it no longer benefits or enjoys less benefits to remain in other foreign markets through this mode as discussed earlier. In other words, Boddewyn [10] found that foreign direct divestment is the opposite of FDI, which entails that before divestment, there must be investment. According to the eclectic paradigm, multi-national corporations engage in foreign direct investment based on three advantages: ownership, location and internalisation advantages.

4.1.2 Strategic motivations of foreign direct investment

Strategic Motivations of Foreign Direct Investment was also adopted by the study. This theory was established by Knickerbocker [12] and later advanced by Graham [13, 14]. The distinguished feature of the strategic approach to FDI is that it believes that an initial inflow of FDI into a country will produce a reaction form the local producers in that country, so that FDI is a dynamic process. Dunning [15] explains that the process from the domestic producers can either be aggressive or defensive in nature. An aggressive response would be a price war or entry into the foreign firm’s home market while a defensive response would be an acquisition or merger of other domestic producers to reinforce market power.

4.2 Empirical review

The influence of trade openness and real gross domestic product (GDP) per capita on FDI is controversial and little or no attention has been paid on FDD. It also seems as if there is no data available for the FDD proxy. However, Chen and Wu [16] add that the determinants of foreign direct investment are also the determinants for foreign direct divestment but with the opposite sign. Therefore, the study will review theoretical and empirical literature on foreign direct investment.

On utmost occasions, it is common that countries seek to attract FDI for several reasons, with various beliefs that FDI allows for more variety of positive societal activities that enable the flow of capital across nations. The variables found to be the most significant determinants of foreign direct investment are openness, market share, return on investment, infrastructure, market size, human capital, real labour costs, exchange rates, political risks, agglomeration and government incentives [2].

Past experiences provide an abundant clarification that FDI encourages exports and allows domestic firms and infant industries to enter the world markets while operating resourcefully by adopting latest technologies and attempt to be competitive, which highlights economic freedom worth to attracting FDI [17] and confirms the cause-and-effect mechanism for exports. Lipsey [18] defines the macroeconomic view as perceiving foreign direct investment as a specific system capital flow across national borders, from home economies to host economies, measured in statistics of balance-of-payments. These flows give rise to a precise form of capital stocks in host economies, precisely the value of home economy investment in entities, such as corporations, regulated by a home-country owner, or where home-country owner has a certain share of voting rights.

The Keynesian theory of investment by Keynes and Fisher as supported by Baddeley [19] and Alchian [20] state that speculations are made until the present estimation of future expected incomes at the margin equal to the opportunity cost of capital. Further, the return on speculation equals to Fisher’s internal rate of return and Keynes’ nominal productivity of capital. The theory highlights the significance of interest rates for investment decisions. A decrease in the interest rates amount to a decrease in the cost of investment in relation to the possible returns. According to this theory, a firm will only invest if the discounted return exceeds the cost of the project. Keynes yet believed that savings do not rely on interest rate but on level of income [21].

Khamis, Mohd and Muhammad [22] attempted to find the influence of inflation rate and GDP per capita on FDI inflows in United Arab Emirates during 1980 to 2013. The study used the ARDL model to examine the long-run relationships between the dependent and independent variables. The findings of the study revealed that inflation has no significant influence on FDI inflows. However GDP per capita proxy used for market size was found to have a significant positive effect on FDI inflows.

There are empirical studies such as those by Edwards [23], Gastanaga, Nugent and Pashamiva [24], Asiedu [25], Na and Lightfoot [26], Cevis and Çamurdan [27], Rogmans and Ebbers [28], Bagli and Adhikary [29], Donghui, Yasin, Zaman and Imran [30] found that FDI was positively related to trade openness of any economy. Musyoka and Ocharo [31] attempted to find the effect of real exchange rate, competitiveness and inflation on foreign direct investment in Kenya using time series data for the 1970 to 2016 period. The study used ordinary least squares regression technique for the variables in study. The findings of the study concluded that competitiveness has a positive and significant influence on FDI inflows, inflation was insignificant for the studied period which concur with the findings of Khamis et al. [22], and lastly real interest rates and exchange rates were found to have a negative significant impact on FDI inflows in Kenya. The study recommended that there is need for favourable interest rates, desirable exchange rates and trade liberalisation over comprehensive programmes to trade reforms, aimed to open the economy and increase its competitiveness, and government must encourage freedom of foreign capital transactions and competition in the local markets.

Kumari and Sharma [2] identified key determinants of FDI inflows in developing countries using unbalanced panel data for the 1990 to 2012 period. The study selected 20 developing countries from the South, East and South-East Asia. Using seven explanatory variables (market size, infrastructure, trade openness, interest rate, inflation, human capital and research and development), the study attempted to find the best fit model from the two models in consideration (fixed effect model and random effect model) with the help of Hausman test. The findings have shown that fixed effect estimation confirms that interest rates, market size, trade openness and human capital yield significant coefficients relative to FDI inflow for the panel of developing countries under study. In addition, findings revealed that market size was the most significant determinant of FDI inflow. The authors recommend than interest rates and inflation must be controlled and monitored since they have an influence on FDI.

Tahmad and Adow [3] investigated the long-run equilibrium relationship of trade openness and foreign direct investment in Sudan by sector during the 1990 to 2017 period. The study used the Johansson co-integration technique and the findings revealed that there is a long-run equilibrium co-integration between trade openness and FDI inflows estimated at negative 0.53 for the aggregate economy when trade openness is measured in terms of the sum of exports and imports over GDP. The degree of openness was estimated at positive values of 0.55, 0.17, and 0.9 for the industrial sector, the aggregate economy and the agricultural sector respectively. The findings indicated that for the studied period, FDI flows for the aggregate economy by sector are influenced by the extent of trade openness in terms of their combined measurement. Furthermore, the greatness of the extent of the industrial trade openness model is a strong one and the government must prioritise this sector regarding exports. The government must also encourage the manufacturing sector, thus promoting attentiveness of FDI in the country’s production sectors and developing infrastructure, particularly those which support the paradigm that Sudan, like various Sub-Saharan African countries, should promote its primary exports to convert from a developing country to a developed one. The study suggests that, according to size of industrial sector trade openness degree, government should use more energy for it to expand and detect this sector as a leading sector utilising trade efficiently and therefore prioritise it in the export.

5. Methodology

This section discusses the methodology used in the study which was inspired by the FDI theory and the interest rates and investment theory. The study adopts an econometric model by using the panel ARDL procedure to examine occurrence of foreign direct divestment in Sub-Saharan countries [32].

5.1 Data

The study used panel annual data of six countries in the Sub-Saharan economies, emerging and developing economies from 1998 to 2018 due to availability of data and these countries were randomly selected to avoid any biases. The selected countries are South Africa, India, Nigeria, Italy, Egypt and Botswana. The secondary data for the following variables: foreign direct investment, trade openness, lending rates, real gross domestic product per capita and urbanisation was obtained from the World Bank.

$$FD{I}_{\mathit{it}}=\propto +{\beta }_1\mathit{LOPENNES}{S}_{\mathit{it}}+{\beta }_{2}L{R}_{\mathit{it}}+{\beta }_3\mathit{LRGD}{P}_{\mathit{it}}+{\beta }_{4}UR{B}_{\mathit{it}}+{\epsilon }_{\mathit{it}}$$E1

where ∝ = represents the constant parameter, LOPENNESS = log of trade openness measured by the sum of imports and exports. Trade openness is expected to have a positive influence on foreign direct investment according to the FDI theory. LR = lending rates which represents the cost of borrowing. Lending rates are expected to have a negative impact on foreign direct investment according to the inverse relationship between interest rates and investment theory. LRGDP = log of real gross domestic product per capita proxy for market size. URB = urbanisation as a percentage of total population.

5.2 Empirical analysis

The following econometric measures are undertaken to investigate the existence of foreign direct divestment in the Sub-Saharan, developing and emerging economies.

5.2.1 The panel unit root test

Before testing for long run cointegration among variables, the study used several tests for stationarity. The panel unit root test is conducted to determine the order of integration among variables which helps in identifying the best suitable model for the data used in the study. The several approaches to unit root testing used in the study for the panel data were Levin, Lin and Chu (2002) (LLC) test; Im, Pesaran, Shin (2003) (IPS) test and the Fisher- Augmented Dickey-Fuller (ADF) test as supported by Maddala and Wu [33].

5.2.2 The panel cointegration test

The panel cointegration test is useful when examining the existence of long run relationships between the regressors and the regressed variables. The Kao panel cointegration test which follows the same basic approach as the Pedroni test extends the Engle-Granger [34] framework to panel cointegration test. The distinct feature of the Kao test from the Pedroni test is that it typically stipulates the cross sections exact intercept and similar coefficients of regressors on the early stage. Also, the Kao and the Pedroni panel cointegration tests are generally used to examine the long run relationship between variables used in a study [35]. The Johansen-Fisher cointegration test uses the findings of the individual independent tests [36]. The Johansen-Fisher panel cointegration pioneered by Maddala and Wu [33] to examine the cointegration in panel data by incorporating the test from individual cross sections to get a test statistic for the whole panel. Say the π_i is the p-value from the distinct cointegration test for cross section i, under the null hypothesis of the panel.

$$-2{\displaystyle \sum _{i-1}^{N}log\left(x^2{\pi }_i\right)\to x^{2}2N}$$E2

Therefore, the value of $x^2$ is built upon the MacKinnon-Haug-Michelis p-values for Johansen’s cointegration trace test and maximum eigenvalue test.

5.2.3 The panel autoregressive distributed lag procedure

The autoregressive distributed lag (ARDL) procedure supported by Pesaran et al. [32] which combines lags of both explained and explanatory variables as regressors is used in the study. The ARDL model is used owing to its ability to join small sample size data and yet generating useful findings [34, 37]. Johansen and Juselius [37] point that the traditional cointegration technique have fewer advantages compared to ARDL that has several advantages. First, it requires small sample size, with variables that are pure I(1), purely I(0) or integrated at different orders of integration but not I(2) [37]. Secondly, it does not require variables to be integrated in the same order compared to the Johansen cointegration approach. Thirdly, ARDL approach caters for any structural breaks in a time series. And lastly, this approach carries a method of measuring the long run and short run findings of one variable on the other and as well distinct both once an appropriate selection of order of the ARDL model is made [38]. Regardless of these advantages, the study employed this model due to its small sample sized panel data and the variables used are integrated at different orders of integration.

6. Results and discussion

Levin, Lin & Chu, Im, Pesaran and Chin W-stat and Augmented Dickey Fuller – Fisher Chi-square tests were employed to perform the panel unit root test and it was discovered that the variables are integrated at different orders of integration $[I\left(0\right)$ and $I\left(1\right)]$ and none of which are $I\left(2\right)$. This gave justification to use panel ARDL. For instance, foreign direct investment was stationary at level, $I\left(0\right)$ for all tests. Trade openness was stationary at $I\left(1\right)$ for IPS and Fisher-ADF. Further, lending rates became stationary at 1%, 5% and 10% level of significance for all tests after first differencing. Gross domestic product was stationary at $I\left(1\right)$ for LLC at 10% level of significance and IPS and Fisher-ADF at 5% respectively. Finally, urbanisation was stationary at $I\left(1\right)$ for all tests (Table 1).

Variables	Tests	Test equations	P-value [level]	P-value [1st difference]
FDI	LLC	Individual & intercept	0.0177	_
		Individual, intercept & trend	0.0113	_
	IPS	Individual & intercept	0.0053	_
		Individual, intercept & trend	0.0812	_
	Fisher-ADF	Individual & intercept	0.0051	_
		Individual, intercept & trend	0.0888	_
LOPENNESS	LLC	Individual & intercept	0.028	0.0022
		Individual, intercept & trend	0.5713	0.1015
	IPS	Individual & intercept	0.0511	0.0011
		Individual, intercept & trend	0.5015	0.0682
	Fisher-ADF	Individual & intercept	0.1048	0.0028
		Individual, intercept & trend	0.5017	0.0996
LR	LLC	Individual & intercept	0.2676	0.0000
		Individual, intercept & trend	0.1315	0.0000
	IPS	Individual & intercept	0.1922	0.0000
		Individual, intercept & trend	0.0743	0.0000
	Fisher-ADF	Individual & intercept	0.1903	0.0000
		Individual, intercept & trend	0.0856	0.0000
LRGDP	LLC	Individual & intercept	0.3227	0.0974
		Individual, intercept & trend	0.4345	0.0782
	IPS	Individual & intercept	0.9084	0.0117
		Individual, intercept & trend	0.6837	0.0309
	Fisher-ADF	Individual & intercept	0.5518	0.0160
		Individual, intercept & trend	0.2319	0.0428
URB	LLC	Individual & intercept	0.0829	0.0000
		Individual, intercept & trend	0.0039	0.0000
	IPS	Individual & intercept	0.9798	0.0000
		Individual, intercept & trend	0.0009	0.0000
	Fisher-ADF	Individual & intercept	0.6535	0.0000
		Individual, intercept & trend	0.0002	0.0007

Table 1.

Summary of panel unit root test results.

Source: Author’s compilation from E-views.

Having established the order of integration for the panel series, the next step is to examine the probability of long-run association between variables. The study will begin with the Kao panel cointegration test. The p-value of 0.004 in the ADF test is less than 0.05 and thus the null hypothesis of no cointegration is rejected and fail to reject the alternative hypothesis of cointegration between the variables. Therefore, the variables have a long run relationship according to the Kao panel cointegration test (Table 2).

Variable	t-Statistic	P-value
ADF	−3.385868	0.0004
Residual variance	3.624602
HAC variance	1.520078

Table 2.

Kao panel cointegration test results.

The test results of the Johansen Fisher panel cointegration with linear deterministic trend test are shown in Table 3. Johansen Fisher panel cointegration test results indicate that the trace statistic has five cointegrating equations and the Fisher maximum-eigen test also shows five cointegrating equations at a 10%, 5% and 1% significance level. The first four equations shows that all p-values are statistically significant at 10%, 5% and 1% level of significance respectively (only one equation at 10%) thus rejecting the null hypothesis of no cointegration. This indicates that there is long run relationship between the variables.

Hypothesised no. of CE(s)	Fisher stat.* (from trace test)	Prob.	Fisher stat.* (from max-eigen test)	Prob.
None	187.9***	0.0000	128.2***	0.0000
At most 1	91.75***	0.0000	57.50***	0.0000
At most 2	44.24***	0.0000	35.63***	0.0004
At most 3	19.57^*	0.0756	18.73^*	0.0953
At most 4	14.56	0.2664	14.56	0.2664

Table 3.

Johansen Fisher panel cointegration with linear deterministic trend test.

Note: *, **, and *** indicate that the p-values are significant at 10%, 5% and 1% level of significance respectively. The Fisher’s test applies regardless of the dependent variable.

In Table 4, the test results of Johansen Fisher panel cointegration with no deterministic trend test indicates that all p-values are significant at 10%, 5% and 1% level of significance respectively (only the last equation at 10%). Therefore, the null hypothesis of no cointegration is rejected, indicating that there is a long run relationship between the variables.

Hypothesised no. of CE(s)	Fisher stat.* (from trace test)	Prob.	Fisher stat.* (from max-eigen test)	Prob.
None	192.5^***	0.0000	105.9^***	0.0000
At most 1	108.7^***	0.0000	62.16^***	0.0000
At most 2	62.58^***	0.0000	40.67^***	0.0001
At most 3	37.08^***	0.0002	32.36^***	0.0012
At most 4	19.05^*	0.0874	19.05^*	0.0874

Table 4.

Johansen Fisher panel cointegration with no deterministic trend test.

Note: *, **, and *** indicate that the p-values are significant at 10%, 5% and 1% level of significance respectively. The Fisher’s test applies regardless of the dependent variable.

Table 5 shows the test results of the Johansen Fisher panel cointegration with Quadratic deterministic trend test. The results indicate that all p-values are significant at 10%, 5% and 1% level of significance, meaning that the null hypothesis of no cointegration is rejected. This indicates that the variables are cointegrated in the long run.

Hypothesised no. of CE(s)	Fisher stat.* (from trace test)	Prob.	Fisher stat.* (from max-eigen test)	Prob.
None	283^***	0.0000	387.1^***	0.0000
At most 1	102.2^***	0.0000	64.2^***	0.0000
At most 2	54.83^***	0.0000	26.8^***	0.0083
At most 3	40.32^***	0.0001	28.75^***	0.0043
At most 4	35.46^***	0.0004	35.46^***	0.0004

Table 5.

Johansen Fisher panel cointegration with quadratic deterministic trend test.

Note: *, **, and *** indicate that the p-values are significant at 10%, 5% and 1% level of significance.

Table 6a and b show the Autoregressive Distributive Lag Short Run and Long Run Results. The long run results indicated that there is an insignificant long run relationship between gross domestic product and FDI (dependent variable) and cannot be used for policy making purposes in this study. These findings contradicts with the findings of Pegkas [39] with FDI as an independent variable. Further, the results showed that lending rates coefficient had a negative significant long run impact on FDI at 5% level of significance. This indicates that if lending rates were to increase by one percent, FDI for the panel six Sub-Saharan and developing economies would decrease by 16.7845%. The findings are in line with Musyoka and Ocharo [31] who discovered that real interest rates have a negative significant impact on foreign direct investment inflows. This further implies that increasing the cost of borrowing will lead to foreign direct divestment in the countries, thus resulting in a decline in inflows. This also indicates that FDI is particularly sensitive to increase in cost of borrowing as suggested by theory on interest rates and investment.

Variable	Coefficient	Std. Error	t-Statistic	P-value
(a) Long run results
LRGDP	−1.411786	8.392655	−0.168217	0.8669
LR	−0.167845	0.080281	−2.090724	0.0400
LOPENNESS	20.6322	1.689844	12.20953	0.0000
URB	−2.236179	0.49482	−4.519181	0.0000
(b) Short run results
ECT(−1)	−0.925834	0.33952	−2.726895	0.0080
D(FDI(−1))	0.131771	0.23861	0.552243	0.5824
D(LRGDP)	−1.231988	32.44146	−0.037976	0.9698
D(LR)	0.308568	0.138535	2.227367	0.0290
D(LOPENNESS)	−4.313282	4.535335	−0.951039	0.3447
D(URB)	−4.99435	2.965234	−1.684302	0.0963
C	69.30302	26.85282	2.580847	0.0118

Table 6.

(a and b) Autoregressive distributed lag short run and long run results.

Notes: D-denotes differenced results for short run.

Trade openness had a positive significant long run relationship with FDI at 10%, 5% and 1% level. This concur with the findings of Kumari and Sharma [2] which revealed that trade openness was significant at 10% level. Furthermore, these findings supports the FDI theory that states that for investment purposes, degree of openness indicates the ease with which a host country is accessible in the world market. Finally, urbanisation showed a negative significant long run relationship with FDI at 10%, 5% and 1% level of significance.

The relationship between the coefficients in Table 6 can also be represented in an equation to further understand their meaning and their influence on FDI. Lending rates and urbanisation both had a significant negative influence on foreign direct investment indicating that an increase in these variables would lead to what we refer to as foreign direct divestment for these selected Sub-Saharan and developing economies. Trade openness showed a positive significant influence on FDI, implying increasing trade openness increases FDI and a decrease leads to foreign direct divestment.

$$\begin{array}{l}FD{I}_{\mathit{it}}=\propto +20.6322\mathit{LOPENNE}{S}_{\mathit{it}}+20.6322L{R}_{\mathit{it}}+-1.4117\mathit{LRGD}{P}_{\mathit{it}}\\ +-2.2362UR{B}_{\mathit{it}}+{\epsilon }_{\mathit{it}}\end{array}$$E3

In the short run, the critical part of the analysis is the error correction term (ECT), which must always be negative according to theory otherwise the model will be explosive and may never return to equilibrium if it is positive. ECT is also referred to the speed of adjustment which shows whether the estimated economic models will be able to return to equilibrium or not and at what speed.

The estimated speed of adjustment, which is at −0.925834, has a negative sign and it is significant at 1% level of significance, as expected by theory. A highly significant speed of adjustment also confirms the existence of cointegration among the variables and a stable long run relationship. This indicates that there is a long-run causality running from the independent variables to the dependent variable and that approximately 93 percent of disequilibrium is corrected each year. It will take 93 percent each year for foreign direct investment activity to return to equilibrium, which is not a slow movement back to equilibrium.

7. Conclusion and recommendations

The aim of the study was to investigate the presence of any noticeable foreign direct divestment, that is any perceptible rapid drop of FDI inflows in the Sub-Saharan African countries using annual panel data from 1998 to 2018. The FDI inflows trends established that there was a rapid decline of FDI inflows for all selected countries after the 2008 global financial crisis and the spillover effects US-China trade war. The study used panel autoregressive distributed lag to determine long run and short run equation for variables that are likely to influence foreign direct investment. The study began by testing for unit root using Levin, Lin & Chu, Im, Pesaran and Chin W-stat and Augmented Dickey Fuller – Fisher Chi-square tests and it was discovered that the variables are integrated at different orders of integration [I(0) and I(1)] and none of which are I(2). The Kao and the Johansen Fisher panel cointegration tests confirmed the long run cointegration among the variables.

The findings of the long run equation revealed that lending rates and urbanisation have a negative and significant influence on foreign direct investment. Further, the findings revealed an insignificant influence of real gross domestic product per capita on FDI. Finally, trade openness showed a positive significant impact on foreign direct investment. This was in line with the FDI theory that states that for investment purposes, degree of openness indicates the ease with which a host country is accessible in the world market. The error correction model also divulged that approximately 93% of disequilibrium will converge towards equilibrium annually.

Like any quantitative or econometrics research, this study also had some limitations. For instance, the period of study ended in 2018 due to unavailability of data for some countries. Lack of data and literature on foreign direct divestment and the absence of data on key variables such as corruption, political risks, labour costs, natural resources and exchange rates may be perceived as limitations. Also, controlling variables such as corruption, exchange rate, political risk and labor cost could make significant improvements to this study. The study also have practical and significant implications for researchers, scholars, governments, policy makers, managers and notably foreign investors.

We recommend policies that increase FDI through cost of borrowing since increasing interest rates result in foreign direct divestment. Real gross domestic product per capita (market size) cannot be used for policy making purposes in the study. Trade openness makes a country accessible on the world market, therefore, policies that promote foreign trade such as exporting complex products, shifting production capabilities from raw materials to more sophisticated products and services, reduce dependence on the primary sector, trade liberalisation, free trade agreements and open trade systems could help reduce the presence of foreign direct divestment in the selected countries. Finally, urbanisation deter foreign direct investment, therefore countries should invest on infrastructure and reduce poverty in rural areas to transform them into urban areas to reduce urbanisation.

JEL classification

E20; F20; F17