SEARCH

SEARCH BY CITATION

Keywords:

  • FDI;
  • exports;
  • Granger causality;
  • error correction;
  • variance decomposition;
  • impulse response;
  • South Africa

Abstract

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests

What started as a financial crisis in 2008–2009 quickly became a crisis of trade for many emerging economies. Using South Africa as a case study, this paper examines the Granger causal relationships between foreign direct investment (FDI), exports and GDP as well as the responsiveness of exports to FDI shocks. The findings indicate that in the long run, FDI has a significant impact on boosting exports. In the short run, there is bi-directional Granger causality between GDP and exports, with uni-directional causality from FDI to exports and FDI to GDP. However, variance decomposition analyses shows that exports are not very responsive to changes in FDI inflow. Copyright © 2012 John Wiley & Sons, Ltd.

1 Introduction

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests

Although most developing countries escaped the worst of the 2008–2009 financial crises, most were heavily hit by the ensuing trade collapse. South Africa is no exception. Between 2008 and 2009, its exports declined by 8.16 per cent (of GDP), whereas its GDP contracted by 1.8 per cent (World Bank, 2011). Foreign direct investment (FDI) inflows around the world also took a heavy knock during the period with global FDI flows declining by 1.03 per cent of global GDP, whereas FDI inflows into sub-Saharan Africa contracted by 0.35 per cent with inflows into South Africa declining by 1.61 per cent of GDP (World Bank, 2011). FDI is highly dependent on the health of the global economy, so in times of crises when investment uncertainty and risk are high, it tends to plummet. Reduced FDI inflows have the potential to result in GDP contractions, with its attendant socio-economic problems. Theoretically, FDI has important spillover effects on the economy and—depending on where it is directed—could boost exports. For instance, Aaron (1999) showed that FDI contributes positively to job creation in the local economy by encouraging investment in human capital through the skills and knowledge transfer to the local workforce by means of specialised training and on-the-job learning. Klein and Olivei (2008) added that FDI inflows also tends to be associated with increased domestic productivity through the transfer of more sophisticated and efficient technology, improved management techniques and enhanced worker training. Global crises such as the recent one, which have the tendency to decrease FDI inflow, are of considerable concern as they could negatively affect exports and consequently economic growth, particularly in developing countries such as South Africa that have an export led growth approach.

In light of the previous discussion, the objective of this paper is twofold: to determine the causality between FDI, exports and GDP in South Africa by using a vector error correction model (VECM) and to analyse the responsiveness of South African exports to shocks in FDI inflow by using variance decomposition and impulse response functions (IRFs). The findings will shed some light on how South Africa and other similar middle-income developing countries can safeguard against reduced FDI inflows that could lead to export and GDP contractions during times of crises. The next section of the paper considers FDI inflow, exports and economic growth in the South African context. An examination of the empirical literature on the FDI–export–GDP nexus is then presented followed by a look at the methodology of the study and data sources. The results of the study are then presented followed by concluding remarks and policy recommendations.

2 Foreign Direct Investment, Exports and Economic Growth in the South African Context

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests

2.1 Exports

South Africa's industrialisation process began prior to the First World War, when the discovery of gold and diamond precipitated the establishment of several related industries such as engineering, explosives manufacturing and cement production. After the end of the First World War, inputs became restricted, forcing South Africa to set up basic industries to produce consumer goods for the domestic market. Exports were very low as a large proportion of what was manufactured was consumed domestically with gold exports fetching the bulk of South Africa's foreign exchange earnings. Fluctuating gold prices and increasing imports created balance of payment problems for the government, who subsequently decided to switch to an export-led growth path. The export-led growth strategy really took centre stage from the 1990s onwards with increased efforts to open up the South African economy further to realise and capitalise on the gains from trade. With the end of the apartheid system and the election of a new government in South Africa in 1994, even more effort was made to liberalise trade and develop the manufacturing and infrastructural base of the economy to boost exports. The Reconstruction and Development Programme set up by the government to define the economic and social agenda explicitly acknowledged that the primary vehicle for achieving increased economic growth was through trade liberalisation and increased competition (Holden, 1996). In accordance with this, a large number of protectionist barriers were phased out in line with WTO rules along with several new initiatives aimed at export promotion. In light of all these initiatives, the 1990s onwards can be characterised as a period that saw a ‘big push’ towards the opening up of South African markets to competition and increased exports and investment. It is thus pertinent to ask if these initiatives have, in a practical sense, contributed to boosting and strengthening the relationship between exports and economic growth.

2.2 Foreign Direct Investment

The imposition of financial and trade sanctions in the early 1980s at the height of apartheid have meant that, historically, South Africa has attracted low levels of FDI. Between 1985 and 1994, average net FDI inflow was −0.38 per cent of GDP primarily because of the many British and American companies that wound down their operations in South Africa during that period (World Bank, 2011). In the last few years, the South African government has devoted considerable effort and resources towards attracting FDI in the hopes of helping to boost economic growth. Theoretically, FDI contributes to economic growth and development by fostering both forward and backward linkages with the domestic economy including providing access to new markets, crowding in domestic investment, promoting increased domestic competition, enhancing productivity and creating employment. FDI tends to agglomerate in sectors and areas already populated by other foreign companies, so increased FDI creates a virtuous cycle of more FDI. Despite the increased efforts to attract FDI into South Africa, the results have been disappointing. Between 1995 and 2008, South African FDI inflows averaged 1.1 per cent of GDP compared with the 3.3 per cent average for other similar middle-income economies (World Bank, 2011). In 2010, the United Nations Conference on Trade and Development inward FDI performance index, which ranks countries on the basis of the FDI they receive relative to the size of their economy, ranked South Africa 103rd out of 141 countries. Arvanitis (2005) attributed this comparatively low FDI inflow to South Africa's low economic growth rates, trade openness, skills shortage and uncompetitive tax rates. To compound matters, positive spillovers from FDI inflows have been limited as they have largely consisted of investments in existing assets with mergers and acquisitions being the primary vehicle. Most of these mergers-type and acquisition-type FDIs are made by foreign investors to gain access to natural resources and regional and local markets with the vast majority of what is produced being sold locally (Jenkins et al., 2000).

3 The Foreign Direct Investment–Export–GDP Nexus

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests

Theoretically, multinational corporations (MNCs) engage in FDI for several reasons, but mostly, these reasons are strategic in nature. These include access to cheap raw materials and other inputs, expansion of existing markets and better service to local clients in the host country. The relationship between FDI and exports can be complementary, substitutory or both (from the home country's perspective). As an example, instead of exporting manufactured goods from the home country to the host country, an MNC could manufacture the goods directly in the host country, reducing export volumes (and costs) from the home country. In this case, the relationship between exports and FDI would be substitutory. On the other hand, the siting of manufacturing operations in the host country could increase imports of intermediate and other inputs as well as related goods and services from the MNC's home country. In this case, the FDI–export relationship would be complementary. Because, theoretically, both of these outcomes are possible, there is an extensive and rapidly expanding literature devoted to the examination of the FDI–export relationship and its effect on the home country's exports. Table 1 highlights some studies that have examined this relationship and their main findings.

Table 1. FDI and exports—effects on the home country's exports
RelationshipMain findingReference
  1. FDI, foreign direct investment.

FDI and exportsFDI tends to reduce exports from the home country (substitution).Helpman (1984); Horst (1972)
FDI tends to boost exports from the home country (complementary)Ekholm et al. (2004); Head and Ries (2001); Camarero and Tamarit (2004)
FDI has both substitutory and complementary effects on home country exportsMarkusen and Maskus (2002); Gray (1998)

Foreign direct investment and export flows also have a direct impact on GDP in terms of the national accounting identity, so there is also extensive literature on the relationship (particularly causality) between FDI and GDP growth on one hand and exports and GDP growth on the other. Table 2 highlights some of these studies. If FDI Granger-causes economic growth in the long run, then we can conclude that the direct and spillover effects of FDI contributes to economic growth. On the other hand, if causality runs in the opposite direction, then it could indicate that investors actively search for new markets with attractive profit-making opportunities. As is obvious from Table 2, the overall results have been mixed. Among the studies reporting bi-directional causality, the results are highly heterogeneous across countries, suggesting that the causality between FDI and economic growth is a country-specific issue, influenced by the economic and technological conditions of the host country.

Table 2. FDI, exports and GDP—Granger causality analyses
RelationshipMain findingReference
  1. FDI, foreign direct investment.

FDI and exportsFDI Granger-causes exportsDritsaki et al. (2004); Cuadros et al. (2004)
Exports Granger-causes FDIMakki and Somwaru, (2004)
Bi-directional Granger causalityLiu et al. (2002)
FDI and GDPFDI Granger-causes GDPZhang (2001); Nair-Reichert and Weinhold (2001)
GDP Granger-causes FDIBasu et al. (2003)
Bi-directional Granger causalityChoe (2003); Chowdhury and Mavrotas (2006)
Exports and GDPExports Granger-causes GDPHenriques and Sardosky (1996); Nidugala (2001)
GDP Granger-causes exportsMbaku (1989); Kormendi and Meguire (1985); De Gregorio (1992); Burney (1996)
Bi-directional Granger causalityGreenaway and Sapsford (1994); Amirkhalkhali and Dar (1995); Sprout and Weaver (1993)

From a review of the empirical literature, a few important issues can be highlighted.

Historically, FDI has flowed from developed to developing countries. Therefore, most the studies analysing the relationship between exports and FDI do so from the perspective of the home country. Very few studies have analysed the FDI–export relationship from the host country perspective, and the majority of these tend to use firm-level and industry-level data. Although this is useful, this study contributes to the literature by using aggregate data to analyse the FDI–export relationship from the host country perspective, providing an important source for comparative analyses. FDI and export flows also have an important effect on GDP through the national accounting identity, so much attention has been focused on the FDI–GDP and GDP–export relationships. These analyses have been carried out largely independently of each other. FDI, exports and GDP are intimately connected, so analysing them separately could introduce omitted variable biases into the analyses. Furthermore, most of the previous studies conducted stationarity tests without accounting for the effect of structural breaks in the data as well as endogeneity between exports and economic growth. This study addresses these shortfalls as indicated in the following discussions.

4 Methodology and Data

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests

4.1 Econometric Model and Data

This study uses VECM Granger causality and impulse response analysis to examine how South African exports respond to FDI shocks. The model tested is as follows:

  • display math(1)

Econometrically,

  • display math(2)

where rEXP denotes seasonally adjusted real exports, rnGDP denotes seasonally adjusted real non-export GDP, FDI denotes foreign direct investment and μ is the independent and identically distributed error term.

Exports are part of GDP in terms of the national accounting identity, so by definition, there will be a positive relationship between exports and economic growth measured in terms of GDP. To overcome this accounting identity problem, real non-export GDP (GDP with exports netted out) is used. Quarterly data for South Africa from the first quarter of 1960 to the fourth quarter of 2009 were used. The data are sourced from the South African Reserve Bank, the primary repository of such data on the South African economy.

4.2 Unit Root Testing

Conventional unit root tests such as the augmented Dickey–Fuller (ADF), Dickey-Fuller Generalised Least Squares (DF-GLS) and Phillips-Perron (PP) tests do not allow for the possibility of structural breaks in the data. Perron (1997) showed that ignoring structural breaks in the data reduces the power of the test to reject a unit root when the stationary alternative is true. Therefore, conventional unit root tests may erroneously reject a stationary series when there are structural breaks in the data. Zivot and Andrews (1992) proposed an alternative unit root test (the ZA test) that takes structural breaks in the data into account when testing for stationarity. The ZA test uses a data-dependent algorithm to endogenously determine the most significant break point in the data on the basis of three models:

  • Model A: This model allows for a one-time break in the level of the series.
    • display math(3)
  • Model B: This model allows for a one-time break in the slope of the trend.
    • display math(4)
  • Model C: This model allows for a one-time break in both the level and slope of the trend function of the series.
    • display math(5)

where DU is the mean-shift dummy variable and DT is the trend-shift dummy variable.

In all models,

  • H0: the series contains a unit root with drift (excluding any structural breaks (α = 0)) and
  • H1: the series is trend stationary with a one-time structural break (α < 1).

The ZA test selects as its break point the data that minimise the one-sided t-statistic for testing inline image, where inline image.

Sen (2003) showed that model C is superior to all other models and hence is the one adopted for this study. The results of the ZA test are presented in Table 3.

Table 3. Zivot–Andrews (ZA) structural break unit root test
VariableLevel [lags]First difference [lags]
 ZA statBreak pointZA statBreak point
  • Notes: ZA stat is calculated from a standard t-distribution.

  • ***

    denotes rejection of the null hypothesis at the 1% level.

Log of rnGDP−2.25[1]1968Q3−6.25[4]***1981Q4
Log of rEXP−3.80[3]1980Q2−12.81[2]***1991Q4
Log of FDI−7.01[8]1999Q1−5.76[11]***1999Q1

4.3 Cointegration Testing

For cointegration testing, the Johansen (1995) full information maximum likelihood test is adopted. This method first establishes the order of the integration of the individual variables and then determines the number of cointegration vectors among the variables to establish their long-run equilibrium relationship. This test is sensitive to the lag length used, so the selection of lag length is critical. This study uses a lag length of 3 selected on the basis of two tests (Appendix A): final prediction error and the Akaike information criterion. An unrestricted intercepts with no trends model (in the VAR and cointegration equation), selected on the basis of the Pantula principle (Johansen, 1992), is used. The results of the cointegration tests are presented in Table 5.

4.4 Vector Error Correction Model Granger Causality Testing

A VECM is used to establish long-run and short-run Granger causality among the cointegrated variables on the basis of the following models:

  • display math(6)
  • display math(7)
  • display math(8)

where ECTt − 1 is the one-period lagged error correction term.

The coefficient on the error correction term captures information on whether past values of the independent variables as a group affect current values of the dependent variable. A significant coefficient would therefore indicate that the independent variables as a group are significant in explaining current values of the dependent variable. The size of the coefficient indicates the speed with which the dependent variable returns to the long-run equilibrium after a disturbance. The coefficients on the individual lagged independent variables capture the short-run dynamics of the system. The results are presented in Table 6.

4.5 Variance Decomposition and Impulse Response Functions

Variance decomposition analysis provides information about the relative importance of innovations to a particular variable in affecting itself and the other variables in the VAR. Variance decomposition analysis therefore sheds light on the relative strength of the relationships among variables by disentangling the portion of the error variance of a variable that is due to its own innovations vis-à-vis innovations due to other variables in the VAR system. IRFs, on the other hand, show how each variable responds to a shock to another variable in the VAR system. Variance decompositions and IRFs are sensitive to the ordering of the variables as the error terms across the VAR system are likely to be correlated to some extent. The residuals are therefore routinely orthogonalised, and different orderings of the variables are tried. However, as Vargas-Silva (2008) pointed out, this approach is only valid if the true model is recursive and only the ordering is unknown, but in most empirical applications, theory provides little guidance as to how the variables should be ordered. A more robust approach would be to use an orthogonalising method that is independent of the ordering of the variables. Pesaran and Shin (1998) provided such an approach—the generalised IRF (GIRF)—which constructs an orthogonal set of innovations that does not depend on the VAR ordering. The GIRF integrates historical patterns of the correlations among different shocks, resulting in unique impulse responses invariant to variable orderings. As such, the response from an innovation to a specific variable is derived by applying a variable-specific Cholesky factor computed with that particular variable at the top of the Cholesky ordering. Given Yt, an n × 1 vector of variables, the VAR moving average model can be represented as

  • display math(9)

inline image is an n × 1 vector of independent and identically distributed errors with a zero mean and variance matrix ∑.

The coefficient matrices Ai can be obtained using the following recursive relation:

  • display math(10)

with A0= In and Ai = 0 for i < 0.

The GIRF is defined to be conditional on only one element at a time, so a jth shock at a time t can be represented as

  • display math(11)

where Ω is the information set. h indicates the time horizon.

So assuming inline image, the effect of a one standard error shock to the jth variable on the expected Yt+h is given by

  • display math(12)

assuming inline image, inline imageand Sj is a selector vector with its jth element equal to unity and zero elsewhere.

5 Results

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests

5.1 Stationarity Tests

All series are non-stationary in levels but stationary when differenced once at the 1 per cent level. The test also identifies structural breaks in each series.

Gross domestic product, exports and FDI all have significant structural breaks at 1981Q4, 1991Q4 and 1999Q1, respectively. The results of the ZA test are confirmed by the ADF test (Table 4), which also suggests that all series are integrated of I(1).

Table 4. Augmented Dickey–Fuller unit root test
VariableLevel[lags]First difference[lags]
 ConstantConstant and trendConstantConstant and trend
  • Notes: Based on MacKinnon (1996) critical values.

  • The lag order was selected from a maximum order of 14 lags by using the Akaike information criterion.

  • ***

    denotes rejection of the null hypothesis at the 1% level.

Log of rnGDP−2.16[4]−2.31[4]−5.52[3]***−5.74[3]***
Log of rEXP−0.83[3]−1.61[3]−12.33[2]***−12.31[2]***
Log of FDI0.89[12]−2.12[8]−4.55[11]***−4.72[11]***

5.2 Cointegration

The trace test indicates the presence of one cointegration vector at the 5 per cent level (Table 5). This implies that GDP, exports and FDI all settle into a stable long-run relationship. However, the maximum eigenvalue test fails to find any evidence of cointegration. Because the trace test is slightly more robust (Liu et al., 2002), the evidence of one cointegration vector is accepted.

Table 5. Unrestricted cointegration rank test (trace and maximum eigenvalue)
Hypothesised no. of CE(s)Trace statistic5% critical valueaMax-eigen statistic5% critical valuea
  • Notes:

  • a

    Mackinnon et al. (1999) p-values.

  • **

    Rejection of the null hypothesis at the 5% level.

  • CE, Cointegrating Equation.

None29.98**29.8014.8821.13
At most 115.0915.5014.7714.26
At most 20.333.840.333.84
5.2.1 Long-run relationship

The VECM suggests the following normalised long-run equilibrium relationship between the variables. This is the primary relation of interest and indicates the long-run elasticities of South African exports with respect to FDI and real GDP.

  • display math(13)

All variables are in natural logs (standard errors in parenthesis).

Foreign direct investment has a significant impact on exports in the long run (a 10 per cent increase in FDI causes a 1.87 per cent increase in exports), underscoring the importance of FDI inflow in South Africa's developmental process. This is consistent with the results of Cuadros et al. (2004) who found a similar positive effect of FDI on exports for a group of Latin American developing economies. In the long run, the effect of GDP in boosting exports is, however, not significant.

5.3 Short-Run Dynamics

The short-run dynamics of the interaction between exports, FDI and GDP can be obtained by estimating an error correction model. The key objective is to ascertain whether exports really drive GDP growth or if it is the other way round. Table 6 presents the results of the error correction models. In each specification, the magnitude of the error correction term indicates the speed of adjustment back to equilibrium. From the results, the coefficients of all the error correction terms are significant. In the equation for exports, the error correction term has a coefficient of 0.095, indicating that 9.5 per cent of disequilibrium is corrected every quarter. Similarly, the error correction coefficient indicates a 23.4 and 4.8 per cent adjustment towards equilibrium each quarter for FDI and GDP, respectively. Columns 3, 4 and 5 show the short-run adjustments of each of the individual variables towards equilibrium in each equation. From these coefficients, Table 7 presents the results of the Granger causality tests.

Table 6. Coefficient estimates of error correction models
Dependent variableECTt−1∑∆LREXP∑∆LFDI∑∆LRNGDP
  • Notes:

  • *

    ,

  • **

    and

  • ***

    indicate rejection of the null hypothesis at the 10%, 5% and 1% significance level, respectively.

LEXP−0.095** (−2.09)0.637 (1.12)0.362** (1.70)0.057* (1.33)
LFDI0.234*** (2.486)0.055 (1.02)−0.210 (−1.08)0.033 (1.06)
LRNGDP0.048*** (2.90)0.892*** (2.42)−0561** (−1.99)0.183 (1.86)
Table 7. Short-run VECM Granger causality tests
H0Wald test/χ2Conclusion
  • Notes:

  • *

    ,

  • **

    and

  • ***

    indicate rejection of the null hypothesis at the 10%, 5% and 1% significance level, respectively.

  • VECM, vector error correction model; FDI, foreign direct investment.

EXP  
FDI does not Granger-cause EXP9.716**Causality
GDP does not Granger-cause EXP6.651*Causality
FDI  
EXP does not Granger-cause FDI1.441No causality
GDP does not Granger-cause FDI2.262No causality
GDP  
FDI does not Granger-cause GDP8.381**Causality
EXP does not Granger-cause GDP11.654***Causality

The results indicate that in the short run, both FDI and GDP play a significant role in spurring exports. This is consistent with theoretical expectations and also with the results of Dristsaki et al. (2004) and Basu et al. (2003) who found such a positive effect of FDI and GDP on exports for other similar developing economies in Africa and Europe. Granger causality therefore runs from FDI to exports and also from FDI to GDP. Exports and GDP, however, have no significant causal effect on increased FDI inflow, but given that the nature of most FDI investment into South Africa is aimed at expanding existing markets with the majority of outputs being sold on the local markets, this is not surprising. FDI investment in the short run may also be influenced by factors such as political stability, market openness, regulatory environment, labour market arrangements, corporate taxes and the level of infrastructural development. With regard to boosting GDP, FDI and exports are both significant in the short run. Specifically, there is evidence of bi-directional causality between exports and GDP, indicating that South Africa's export-led growth approach may be bearing fruits.

5.4 Variance Decomposition

5.4.1 Exports

Even in the 20th period, 88 per cent of the variations in exports are explained by its own innovations, with 3.58 and 8.51 per cent explained by variations in FDI and GDP, respectively (Table 8). This implies that shocks to export volumes tend to persist for a long time. Exports are also not very responsive to FDI and GDP up to and possibly beyond the 20th period.

Table 8. Variance decomposition of South African exports
Variance decomposition of LREXP
PeriodS.E.LREXPLFDILRNGDP
10.053900100.00000.0000000.000000
40.06782993.357682.4557874.186529
80.08296791.369732.0847216.545546
120.09373789.907142.3515767.741281
160.10223388.830772.8902708.278962
200.10932087.915703.5764058.507900
5.4.2 Foreign direct investment

In the first period, 99.9 per cent of variation in FDI is explained by FDI innovation. By the 20th period, this drops to 83 per cent, with 15.92 and 1.20 per cent explained by variations in exports and GDP, respectively (Table 9). The influence of exports on FDI is thus bigger than the influence of FDI on exports. This is not surprising as increases in exports tend to be correlated with other factors such as improved terms of trade and an improved regulatory and tax environment, which tends to attracts even more FDI. In contrast, the influence of FDI on exports is mostly realised only after spillover effects have become entrenched.

Table 9. Variance decomposition of South African FDI
Variance decomposition of LFDI
PeriodS.E.LREXPLFDILRNGDP
  1. FDI, foreign direct investment.

10.1117760.05222199.947780.000000
40.1909801.67624697.110681.213078
80.2501934.38934493.786111.824550
120.2948207.96706690.324221.708717
160.33250011.9164886.641861.441660
200.36628315.9156982.884851.199460
5.4.3 GDP

The variance decomposition of GDP is quite interesting as it shows that shocks to non-export GDP have an immediate and significant impact on exports—exports explain 48 per cent of GDP variation. However, this steadily declines to only 5 per cent by the 20th period. This underscores the importance that export growth plays in spurring GDP growth in South Africa. FDI is somewhat important in the long run as well, explaining roughly 6 per cent of the variation in GDP in the 20th period (Table 10).

Table 10. Variance decomposition of non-export GDP
Variance decomposition of LRNGDP
PeriodS.E.LREXPLFDILRNGDP
10.01976347.796730.03170552.17156
40.03284321.869001.53534176.59566
80.04992010.086392.28893287.62468
120.0654896.1039373.74069490.15537
160.0802934.9707145.17672789.85256
200.0945615.0221236.43340988.54447
5.4.4 Export Response to foreign direct investment shock

Figure 1 shows the response of exports to a one standard deviation innovation in FDI. A one standard deviation innovation in FDI in one quarter causes an immediate drop in exports. This negative response persists into the second quarter, after which exports begin to respond positively, returning to pre-shock levels in the third quarter. Exports increase further in the fourth quarter and somewhat stabilises from the eighth quarter onwards. This positive response persists beyond the 20th quarter. What we can deduce from this is that FDI innovations do not seem to have a positive effect on exports in the first three quarters. This may be reasonable for two reasons. Firstly, most FDI into South Africa has traditionally not been geared towards expanding export markets, so the initial influence on exports might be negligible. In the long run, however, spillover benefits may be realised in export sectors, accounting for the new higher export equilibrium beyond the eighth quarter. Secondly, even if the initial FDI outlays were directed towards boosting exports, it takes time to break into the market and establish a foothold, so initially, the effect may be negligible with noticeable positive effects on exports only being realised from the fourth quarter onwards. Additionally, what this tells us is that negative shocks to FDI may negatively affect exports, persisting into the medium-term to long-term unless positive inventions are made.

image

Figure 1. Export response to one standard deviation innovation in foreign direct investment

Download figure to PowerPoint

6 Concluding Remarks and Policy Implications

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests

This paper has examined the Granger causal relationships between South African FDI, exports and GDP as well as the responsiveness of exports to FDI shocks. The findings indicate that in the long run, FDI has a significant impact on boosting exports with a 10 per cent increase in FDI inflow, resulting in a potential 1.87 per cent increase in export volumes. In the short run, there is bi-directional Granger causality between GDP and exports, with uni-directional causality from FDI to exports and FDI to GDP. This highlights the importance of FDI to South Africa's export-led growth strategy. However, variance decomposition analysis shows that exports are not very responsive to changes in FDI inflow. This may be due to the nature of South African FDI inflows, which are primarily through mergers and acquisitions and directed towards production of goods and services for the local market. South Africa's FDI inflows are also comparatively low possibly because of low economic growth rates, skills shortages and uncompetitive tax rates. For policy makers, this presents an opportunity to introduce remedial measures that will address the aforementioned issues. Crucially, it is important not only that South Africa attracts FDI but also that it attracts the right kind of FDI—those that will boost its exports. Ensuring that South Africa attracts the right kind of FDI in appropriate sectors will complement its export-led growth approach.

Variance decomposition analyses also show that shocks to GDP have an immediate and significant impact on South African exports. Although this poses considerable risks to exports in the event of recessionary episodes, it is an opportunity for policy makers to take measures to ensure that any sudden unpredictable contractions in GDP are brief. These measures could include streamlining the legislative and regulatory processes required for government and the monetary authorities to respond to future crises in order to speed up the process. Impulse response analyses indicate that although the initial response in the first two periods is negative, exports tend to respond positively to innovations in FDI beyond the third quarter, settling into a somewhat stable higher new equilibrium beyond the eighth quarter. Although this may be good news, it also underscores the risks associated with sudden drops in FDI inflow. In light of the recent financial crises, which caused a marked drop in South African FDI inflow, there is an urgent need for policy makers to do more to attract and safeguard FDI inflows.

References

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests
  • Aaron C. 1999. The contribution of FDI to poverty alleviation. Report from the Foreign Investment Advisory Service presented at a conference in Singapore.
  • Amirkhalkhali S, Dar AA. 1995. A varying-coefficients model of export expansion, factor accumulation and economic growth: evidence from cross-country, time series data. Economic Modelling 12(4): 435441.
  • Arvanitis A.. 2005. Foreign direct investment in South Africa: why has it been so low? In Post-Apartheid South Africa: The First Ten Years, Nowak M, Riccil A (eds). Washington D.C: International Monetary Fund.
  • Basu P, Chakraborty C, Reagle D. 2003. Liberalization, FDI, and growth in developing countries: a panel cointegration approaches. Economic Inquiry 41(3): 510516.
  • Burney NA. 1996. Exports and economic growth: evidence from cross country analysis. Applied Economics Letters 40(3): 36973.
  • Camarero M, Tamarit C. 2004. Estimating the export and import demand for manufactured goods: the role of FDI. Review of World Economics 140(3): 347375.
  • Choe JI. 2003. Do foreign direct investment and gross domestic investment promote economic growth? Review of Development Economics 7(3): 4457.
  • Chowdhury A, Mavrotas G. 2006. FDI and growth: what causes what? World Economy 29(1): 919.
  • Cuadros A, Orts V, Alguacil M. 2004. Openness and growth: re- examining foreign direct investment, trade and output linkages in Latin America. Journal of Development Studies 40(4): 167192.
  • De Gregorio J.. 1992. Economic growth in Latin America. Journal of Development Economics 39(1): 5984.
  • Dritsaki M, Dritsaki C, Adamopoulos A. 2004. A causal relationship between trade, foreign direct investment, and economic growth for Greece. American Journal of Applied Sciences 1(3): 230235.
  • Ekholm K, Forslid R, Markusen JR. 2004. Export-platform foreign direct investment. IIIS Discussion Paper No. 50, Dublin: Institute for International Integration Studies.
  • Gray HP. 1998. International trade and foreign direct investment: the interface. In Globalization, Trade and Foreign Direct Investment, J. H. Dunning (ed). Elsevier: Oxford; 1927.
  • Greenaway D, Sapsford D. 1994. Exports, growth, and liberalization: an evaluation. Journal of Policy Modelling 16(1): 16586.
  • Head K, Ries J. 2001. Overseas investment and firm exports. Review of International Economics 9(1): 108122.
  • Helpman E. 1984. A Simple Theory of International Trade with Multinational Corporations. Journal of Political Economy 92(3): 134142.
  • Henriques I, Sardosky P. 1996. Growth or Growth in the Global Ambridge. MIT Press: MA.
  • Holden M. 1996. Economic Integration and Trade Liberalization in Southern Africa. World Bank Discussion Paper No. 342: Washington, DC.
  • Horst T. 1972. The industrial composition of U.S. exports and subsidiary sales to the Canadian market. American Economic Review 62(1/2): 3745.
  • Jenkins C, Leape J, Thomas L. 2000. Gaining from trade in Southern Africa: complementary policies to underpin the SADC Free Trade Area. Commonwealth Secretariat/Macmillan.
  • Johansen S. 1992. Determination of cointegration rank in the presence of a linear trend. Oxford Bulletin of Economics and Statistics 54(2): 383397.
  • Johansen S. 1995. Likelihood-based Inference in Cointegrated Vector Autoregressive Models. Oxford University Press.
  • Klein M, Olivei G. 2008. Capital account liberalization, financial depth, and economic growth. Journal of International Money and Finance 27(6): 861875.
  • Kormendi RC, Meguire PG. 1985. Macroeconomic determinants of growth: cross country evidence. Journal of Monetary Economics 16(1): 14163.
  • Liu X, Burridge P, Sinclair PJN. 2002. Relationship between economic growth, foreign direct investment and trade: evidence from China. Applied Economics 34(1): 14331440.
  • Mackinnon JG. 1996. Numerical distribution functions for unit root and cointegration tests. Journal of Applied Econometrics 11(6): 601618.
  • Mackinnon JG, Haug A, Michelis L. 1999. Numerical distribution functions of likelihood ratio tests for cointegration. Journal of Applied Econometrics 14(1): 563-577.
  • Makki SS, Somwaru A. 2004. Impact of Foreign direct investment and trade on economic growth: evidence from developing countries. American Journal of Agricultural Economics 86(3): 795801.
    Direct Link:
  • Markusen JR, Maskus KE. 2002. Discriminating among alternative theories of the multinational enterprise. Review of International Economics 10(1): 694707.
  • Mbaku JM. 1989. Export growth and economic performance in developing countries: further evidence from Africa. Journal of Economic Development 14(1): 12742.
  • Nair-Reichert U, Weinhold D. 2001. Causality tests for cross-country panels: a new look on FDI and economic growth in developing countries. Oxford Bulletin of Economics and Statistics 63(1):153171.
  • Nidugala GK. 2001. Exports and economic growth in India: an empirical investigation. Indian Economic Journal 47(3): 6778.
  • Perron P.. 1997. Further evidence on breaking trend functions in macroeconomic variables. Journal of Econometrics 80(1): 355385.
  • Pesaran MH, Shin Y. 1998. Generalised impulse response analysis in linear multivariate models. Economics Letters 58(1): 1729.
  • Sen A.. 2003. On unit root tests when the alternative is a trend break stationary process. Journal of Business and Economic Statistics 21: 174184.
  • Sprout RVA, Weaver JH. 1993. Exports and economic growth in a simultaneous equations model. Journal of Developing Areas 27(2): 289306.
  • Vargas-Silva C. 2008. Are remittances manna from heaven? a look at business cycle properties of remittances. The North American Journal of Economics and Finance 19(3): 290303.
  • World Bank. 2011. World Development Indicators. Washington, DC.
  • Zhang KH. 2001. Does foreign direct investment promote economic growth? Evidence from East Asia and Latin America. Contemporary Economic Policy 19(2): 175185.
  • Zivot E, Andrews D. 1992. Further evidence of great crash, the oil price shock and unit root hypothesis. Journal of Business and Economic Statistics 10(3): 251270.

Appendix A: Var Lag Length Selection Tests

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Foreign Direct Investment, Exports and Economic Growth in the South African Context
  5. 3 The Foreign Direct Investment–Export–GDP Nexus
  6. 4 Methodology and Data
  7. 5 Results
  8. 6 Concluding Remarks and Policy Implications
  9. References
  10. Appendix A: Var Lag Length Selection Tests
LagFinal prediction error testAkaike information criterion
00.0004970.905793
19.94e−09−9.912684
28.97e−09−10.01637
38.90e−09*−10.02386*
49.31e−09−9.979750
59.33e−09−9.977914
69.82e−09−9.927626
71.05e−08−9.864672
81.02e−08−9.894336
99.72e−09−9.942445
101.01e−08−9.902947
111.09e−08−9.833807
121.18e−08−9.759836
131.18e−08−9.765408
141.23e−08−9.726661

Note: * indicates lag order selected by the criterion.