Volume 27 Issue 3 Article 1 September 2025 The Effect of Disaggregated Country Risk on Foreign Portfolio The Effect of Disaggregated Country Risk on Foreign Portfolio Investment Flows in South Africa Investment Flows in South Africa Paul-Francois Muzindutsi University of KwaZulu-Natal, School of Accounting, Economics, and Finance, South Africa, MuzindutsiP@ukzn.ac.za Tristan Kyle Govender University of KwaZulu-Natal, School of Accounting, Economics, and Finance, South Africa Nokwanda Nkwanyana University of KwaZulu-Natal, School of Accounting, Economics, and Finance, South Africa Sanelisiwe Zulu University of KwaZulu-Natal, School of Accounting, Economics, and Finance, South Africa Nondumiso Myeni University of KwaZulu-Natal, School of Accounting, Economics, and Finance, South Africa See next page for additional authors Follow this and additional works at: https://www.ebrjournal.net/home Part of the Finance Commons, Finance and Financial Management Commons, and the Portfolio and Security Analysis Commons Recommended Citation Recommended Citation Muzindutsi, P ., Govender, T., Nkwanyana, N., Zulu, S., Myeni, N., Khuzwayo, S., & Dube, F. (2025). The Effect of Disaggregated Country Risk on Foreign Portfolio Investment Flows in South Africa. Economic and Business Review, 27(3), 130-140. https://doi.org/10.15458/2335-4216.1356 This Original Article is brought to you for free and open access by Economic and Business Review. It has been accepted for inclusion in Economic and Business Review by an authorized editor of Economic and Business Review. The Effect of Disaggregated Country Risk on Foreign Portfolio Investment Flows The Effect of Disaggregated Country Risk on Foreign Portfolio Investment Flows in South Africa in South Africa Authors Authors Paul-Francois Muzindutsi, Tristan Kyle Govender, Nokwanda Nkwanyana, Sanelisiwe Zulu, Nondumiso Myeni, Sinegugu Khuzwayo, and Fikile Dube This original article is available in Economic and Business Review: https://www.ebrjournal.net/home/vol27/iss3/1 ORIGINAL ARTICLE The Effect of Disaggregated Country Risk on Foreign Portfolio Investment Flows in South Africa Paul-FrancoisMuzindutsi a, * ,TristanKyleGovender a ,NokwandaNkwanyana a , SanelisiweZulu a ,NondumisoMyeni a ,SineguguKhuzwayo a ,FikileDube b a University of KwaZulu-Natal, School of Accounting, Economics, and Finance, South Africa b University of the Western Cape, Faculty of Economic and Management Sciences, South Africa Abstract This study explores the relationship between disaggregated country risk and foreign portfolio investment (FPI) ows in South Africa, focusing on both the long-run and short-run effects of economic, nancial, and political country risk measures on net foreign purchases of shares (NFPS) and net foreign purchases of bonds (NFPB) during the period from 1995 to 2019. We employed autoregressive distributed lag (ARDL) and nonlinear autoregressive distributed lag (NARDL) models to assess the relationships between the variables. The results indicate that all disaggregated country risk measures have a long-run effect on NFPS and NFPB, and the impacts of these risks are asymmetric. Specically, low levels of economic risk are associated with a decline in foreign equity ows and an increase in foreign bond investments in the long run, while high levels of economic risk correlate with a rise in both foreign equity and bond investment ows. Conversely, both high and low levels of nancial and political risk lead to a decrease in NFPS and NFPB. Notably, nancial risk was the only country risk measure found to signicantly impact NFPB in the short run. The ndings highlight the importance for policymakers to understand these complex relationships in order to implement strategies that foster a mutually benecial economic, political, and nancial climate in South Africa, encouraging FPI while maintaining sovereignty. Keywords: Country risk, Political risk, Financial risk, Economic risk, Net foreign portfolio investment JEL classication: E44, G10, G11 1 Introduction F or a developing country such as South Africa, foreign capital inows in the form of foreign direct investment (FDI) and foreign portfolio invest- ment (FPI) play a critical role in promoting economic growth (Chorn & Siek, 2017). FPI investors are consid- ered passive, as they do not participate in the day-to- day operations of domestic corporations, while FDI investors are more actively involved. Following the political liberalisation after 1994, South Africa became a signicant recipient of FPI (Bah & Giritli, 2020). From 2013 to 2018, South Africa was the leading re- cipient of foreign portfolio investment in sub-Saharan Africa (Schwab, 2017). During this period, FPI inows amounted to R408.1 billion, three times greater than FDI inows (OECD, 2018). However, data from The Global Economy indicates a signicant drop in FPI from 2019 to 2021, which relegates South Africa to the lowest recipient of FPI in the region. This transition raises critical questions about the factors inuencing FPI inows. Were there changes in the characteris- tics of the South African economy that deterred FPI investors? While portfolio investments are considered less risky due to their liquidity, this same liquidity poses a risk for the receiving country, as investors may withdraw their investments during periods of nan- cial, political, or economic instability (Al Samman & GabAlla, 2020; Bah & Giritli, 2020). These polit- ical, nancial, and economic risk factors constitute Received 8 April 2024; accepted 18 June 2025. Available online 10 September 2025 * Corresponding author. E-mail address: MuzindutsiP@ukzn.ac.za (P .-F. Muzindutsi). https://doi.org/10.15458/2335-4216.1356 2335-4216/© 2025 School of Economics and Business University of Ljubljana. This is an open access article under the CC BY license (http://creativecommons.org/ licenses/by/4.0/). ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 131 country risk, which aligns with Nhlapho and Muzin- dutsi (2020), who assert that capital inows and outows are inuenced by country risk. The past decade in South Africa has been tumultuous, marked by issues of state capture, credit rating downgrades, declining socio-economic conditions, and high na- tional debt. Political instability peaked in 2016 during the governmental elections when the African Na- tional Congress (ANC) lost support in four major metropolitan cities, which led to increased political uncertainty, cabinet reshufes, and votes of no con- dence against the previous and current presidents (World Bank Group, 2018). Additionally, rising inter- est rates, high ination, risk rating downgrades, load shedding, and civil unrest have contributed to consis- tently low economic growth (National Treasury, 2022; Nhlapho & Muzindutsi, 2020). Do these components of country risk affect foreign portfolio investment? This study seeks to answer this question. Most existing literature on South African invest- ments focuses on FDI, leaving a gap in understanding FPI and its relationship with country risk. Using au- toregressive distributed lag (ARDL) and nonlinear autoregressive distributed lag (NARDL) models, this study examines the short-run and long-run inuences of political, nancial, and economic risk on FPI ows in South Africa. The contributions of this study to the literature are fourfold. First, it analyses the in- uences of disaggregated country risk on FPI ows in a developing country context, a topic that has not been previously explored. Second, most studies that assess the relationship between FPI ows and disaggregated country risk focus on developed coun- tries, with limited attention to emerging economies. Emerging economies have diverse nancial markets, levels of economic development, cultural factors, and geographical contexts, all of which signicantly inu- ence foreign capital ows. As such, models developed for one economy may not yield accurate results when applied to another. Thus, it is essential to investigate the effects of disaggregated country political, nan- cial, and economic risk on FPI ows, specically for South Africa. Third, South African literature often em- phasises political risk while neglecting nancial and economic risk components. Nhlapho and Muzindutsi (2020) suggested that greater attention should be paid to the effects of disaggregated country risk on for- eign nancial ows. Fourth, existing studies suggest that the debate on the inuence of country risk on FPI ows remains unresolved and warrants further investigation. 2 Literature review The risk–return trade-off is a fundamental princi- ple in nance that suggests higher potential returns are generally associated with higher levels of risk (Elton et al., 2009). This phenomenon is explained by modern portfolio theory (MPT), which posits that investors can optimise returns by constructing diversied portfolios that balance risk and return. From a global point of view, foreign net portfolio investments are often guided by MPT principles, as international investors seek exposure to uncorrelated or higher-yielding assets in emerging or frontier mar- kets (Zaimovic et al., 2021). However, linking the country risk and the concept of risk–return trade-off would emphasise how the risks associated with a particular country inuence foreign investment deci- sions (Hassan, 2023). Thus, foreign investments tend to ow from countries with higher risk or lower re- turns to those with more stable environments and higher returns as investors seek to maximise returns while minimising risk (Al Samman & GabAlla, 2020). In other words, shifts in risk perceptions or economic conditions inuence the direction and volume of for- eign investment ows across countries, and country risk components play a crucial role in these dynamics. Thus, these dynamics inform the theoretical link be- tween country risk components and foreign portfolio investment inows. Country risk arises from political, economic, and social issues, representing the overall risk landscape of a nation. Assessing country risk can inuence the creation of an optimal investment portfolio, where a weighted composite indicator is tailored to a spe- cic risk level (Malala & Adachi, 2020). Typically, a risk-averse investor may be deterred from investing in economies with high country risk levels. Con- versely, risk-takers might pursue investments in such environments due to the potential for high returns as- sociated with elevated risk (McCue, 2000). Therefore, the effect of country risk on foreign capital inows can be either positive or negative. In South Africa, the ow of foreign capital largely depends on whether in- vestments offer sufcient returns to offset the inherent risks of the economy (Oleksiv, 2000). Despite its importance, only a few studies focus on the specic components of country risk when evalu- ating their inuence on foreign nancial ows, such as FDI and FPI. Most existing literature (e.g., Hassan, 2023; Hayakawa et al., 2013; Sekkat & Veganzones- Varoudakis, 2007; van Wyk & Lal, 2008) has primar- ily analysed the impact of macroeconomic factors, such as GDP , exchange rates, and ination, on FPI, with a greater emphasis on FDI. For instance, Topal and Gül (2016) examined data from 49 developing countries and found that nancial risk does not sig- nicantly inuence FDI inows; however, decreases in economic and political risk positively affect FDI. Similarly, Rafat and Farahani (2019) studied the re- lationship between FDI and political risk in Iran 132 ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 and discovered a negative impact of political risk on FDI. Using a panel dataset covering 72 nations, Sekkat and Veganzones-Varoudakis (2007) suggested that higher GDP indicates better market opportunities for attracting FDI and that political and nancial risks signicantly inuence FDI. Their ndings also re- vealed that higher risk levels lead to decreased FDI. Additionally, Hayakawa et al. (2013) analysed FDI inows for 89 countries from 1985 to 2007, concluding that political risk negatively correlates with FDI. Their results indicated that initially, low political risk and a subsequent decrease in political risk attract more FDI inows, while emerging economies with reduced - nancial risk do not necessarily attract FDI, suggesting that the sensitivity of FDI to country risk components is not uniform. In South Africa, Magoane et al. (2023) evaluated the relationships between political risk, the exchange rate, and FDI, nding that these factors inuence FDI in the long run. However, they noted that the real exchange rate and political risk do not signicantly af- fect FDI in the short run, whereas GDP does. Similarly, Khan and Akbar (2013) identied a long-run asso- ciation between FDI and political risk, while Meyer and Habanabakize (2018) found that both economic growth and political risk inuence FDI levels over the short and long term. Their ndings highlighted that political risk ratings have a more considerable effect on FDI ows than GDP , emphasising the importance of specic country risk components. Regarding FPI, Al Samman and GabAlla (2020) explored the effects of country risk components on foreign equity portfolio inows, identifying a long- run association between net foreign portfolio equity investment inows and all three country risk mea- sures. Specically, their results indicated that political risk inuences inows positively, while economic and nancial risks affect them negatively. Mutize and Gos- sel (2019) investigated the net inuence of sovereign credit rating announcements on foreign-currency- denominated stock and bond markets across 19 African economies from 1994 to 2014. They found weak positive relationships between bond and stock markets and sovereign credit ratings, suggesting that these markets respond positively to credit upgrades and negatively to downgrades. Furthermore, they re- vealed that variations in sovereign ratings have a more pronounced effect on bond prices than stock prices, indicating that the impact of country risk components may differ between bond and stock markets. Overall, existing studies primarily focus on the inuence of country risk on FDI ows, with limited at- tention to its effects on FPI. Given that FPI inows are gaining prominence for enhancing investment ows in host countries (Singhania & Saini, 2018), assessing how changes in country risk impact FPI dynamics is crucial. This study aims to contribute to this discourse by analysing the effects of disaggregated country risk on FPI ows in South Africa, explicitly examining net foreign purchases of shares (NFPS) and net for- eign purchases of bonds (NFPB). Previous research has indicated that the performance of stock and bond markets inuences foreign portfolio investments pos- itively (Haider et al., 2017; Singhania & Saini, 2018). However, these markets are also susceptible to coun- try risk, affecting foreign nancial ows. The rela- tionship between market performance, asset returns, and country risk is complex. In South Africa, Nhlapho and Muzindutsi (2020) found that political risk affects both bond and stock returns in the short and long run, while economic risk has short-run effects on bond returns. Muzindutsi and Obalade (2024) discovered that bond returns increase with rising political and economic risks but decrease as nancial risk increases. These conicting ndings on the effects of country risk components on bond and stock markets neces- sitate further investigation to clarify this important topic. The theories and reviewed empirical evidence re- vealed that minimising country risk is crucial for countries seeking to attract foreign capital, as higher risks tend to deter foreign investors, affecting the net inows of foreign investment. In contrast, high risk may attract foreign investors with high level of risk appetite. Thus, this study has the following three prior expectations or hypotheses. First, we hypothe- sise that increasing country risk (economic, nancial, and political risks) decreases the FPI inows. This expectation is based on the traditional nance theory that investors are less likely to invest in countries perceived as high risk, leading to reduced foreign investment inows. However, behavioural nance argues that behavioural biases such as the level of risk appetite or tolerance and investor irrational be- haviours may change this phenomenon (Mittal, 2022), where investors may channel their investments in countries in search of high returns (Baek & Qian, 2011). Hence, the second prior expectation is a pos- itive relationship between country risk components and net foreign portfolio ow. Lastly, investors have been found to react to political risk as opposed to nancial and economic risks, as the level of risk tol- erance across economic, political, and nancial risks may differ. Thus, our third hypothesis is that the ef- fect of economic, nancial, and political components of country risk on net foreign portfolio ows varies because the investors may react differently to each of these risk components. Thus, the empirical analysis of this study is set to examine these three prior expecta- tions/hypotheses. ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 133 3 Data and methodology 3.1 Data and variables This analysis employs a quantitative approach, utilising time-series data to achieve its objectives. Monthly data on NFPS and NFPB from January 1995 to December 2019 are used in this study. This sample period was selected because economic sanctions on South Africa ended in 1995, marking the transition to a democratic regime. The end date of 2019 was chosen to exclude the COVID-19 pandemic, which caused unprecedented disruptions in the global econ- omy and affected investment ows. The NFPS and NFPB data, sourced from the South African Re- serve Bank, reect the overall market value of South African stocks and bonds purchased by non-resident investors minus the stocks and bonds sold (Have- mann et al., 2022). NFPS and NFPB are appropriate measures of FPI ows, as evidenced by their use in several studies (Mamvura & Sibanda, 2020; Muguto et al., 2019). Data on country risk ratings was obtained from the International Country Risk Guide (ICRG) scores provided by the PRS Group. The ICRG comprises 22 variables categorised into three subgroups of country risk: nancial risk, political risk, and economic risk. Political risk encompasses factors such as government stability, socio-economic conditions, corruption, in- ternal and external conicts, military inuence in pol- itics, religious tensions, bureaucratic quality, ethnic tensions, law and order, and democratic accountabil- ity. Economic risk includes indicators such as GDP per capita, real GDP growth, annual ination rate, budget balance as a percentage of GDP , and current account balance as a percentage of GDP . Financial risk consists of foreign debt as a percentage of GDP , foreign debt service as a percentage of exports, current account balance as a percentage of exports, net international liquidity measured in months of import cover, and exchange rate stability. Table 1 summarises the risk rating scores. The political risk index is considered the most critical component of country risk and is weighted at 100 points, while nancial and economic risk indices are assigned weights of 50 points each Table 1. Country risk rating scores. Financial risk and Risk rating score Political risk economic risk Very low 80.0%–100% 40.0%–50% Low 70.0%–79.9% 35.0%–39.9% Moderate 60.0%–69.9% 30.0%–34.9% High 50.0%–59.9% 25.0%–29.9% Very high 0.0%–49.9% 0.0%–24.9% Source: The PRS Group (2022). (Muzindutsi et al., 2022). The overall scores from these three indices are averaged to derive a comprehensive country risk score. The scores range from 0 to 100, classied as follows: very high risk (0 to 49.9 points), high risk (50 to 59.9 points), moderate risk (60 to 69.9 points), low risk (70 to 79.9 points), and very low risk (80 to 100 points; The PRS Group, 2022). 3.2 Model specication In order to achieve the aim of this study, which in- volves assessing the effect country risk measures have on net foreign ows and whether that effect is asym- metric or not, the linear ADRL and nonlinear ARDL (NARDL) models were adopted. These models are su- perior to other cointegration models such as Johansen and Engle–Granger tests because of their simplicity and their ability to produce accurate estimates even if the series is integrated of different orders (Allen & McAleer, 2021; Muzindutsi & Mjeso, 2018). This advantage is benecial for the present study as the dataset used is more likely to be integrated of dif- ferent orders. In determining the linear relationship between disaggregated country risk and FPI ows in South Africa, the following linear ARDL models were employed: 1NFPS t Da 0 C n X iD1 b i 1NFPS t i C n X iD0 c i 1LER t i C n X iD0 d i 1LFR t i C n X iD0 e i 1LPR t i C0 1 NFPS t 1 C0 2 LER t 1 C0 3 LFR t 1 C0 4 LPR t 1 C+ 1t (1) and 1NFPB t Da 0 C n X iD1 b i 1NFPB t i C n X iD0 c i 1LER t i C n X iD0 d i 1LFR t i C n X iD0 e i 1LPR t i C0 1 lny t 1 C0 2 LER t 1 C0 3 LFR t 1 C0 4 LPR t 1 C+ 1t (2) Eqs. (1) and (2) represent the change in NFPS and NFPB, respectively. The NFPS and NFPB equations are approximated independently. LER, LFR, and LPR represent the natural logarithms for economic risk, nancial risk, and political risk, respectively. The nat- ural logarithm is used to transform the highly skewed country risk variable scores (see Table 2) into vari- ables that approximate a normal distribution (Benoît & Dubra, 2011). Short-run coefcients are represented by a, b, c, d, and e, while 0 i represent the long-run coefcients, and + denotes the error term. To estimate the long-run relationship between disaggregated 134 ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 country risk and FPI ows, the following hypotheses are tested: H 0 :0 1 ;0 2 ;0 3 ;0 4 D 0 H 1: 0 1 ;0 2 ;0 3 ;0 4 6D 0 The null hypothesis expresses that there is no long- run relationship association (cointegration) between the series, whereas the alternate hypothesis expresses that there is a long-run association between the se- ries. If there is conrmation of a long-run relationship between disaggregated country risk and FPI ows, then error correction models (ECM) are estimated to evaluate the short-run relationships. The ECMs are expressed as follows: 1NFPS t Da 0 C k X iD1 b i 1NFPS t i C k X iD0 c 1i 1LER t i C k X iD0 d 1i 1LFR t i C k X iD0 e 1i 1LPR t i Cd+ t 1 C u t (3) and 1NFPB t Da 0 C k X iD1 b i 1NFPB t i C k X iD0 c 1i 1LER t i C k X iD0 d 1i 1LFR t i C k X iD0 e 1i 1LPR t i Cd+ t 1 C u t (4) Conceptually,d is the error correction term that cap- tures the adjustment speed to the equilibrium. Due to the possibility of nonlinearity in the relationship, the NARDL model is used to supplement the ARDL model. The NARDL model expands the linear ARDL model by considering the possibility of a nonlinear association between the variables. The NARDL model breaks down the variables in the model into posi- tive and negative changes to identify whether there is a nonlinear effect on the dependent variable (Kartal et al., 2022). The NARDL model designed by Shin et al. (2011) is implemented to approximate both short and long-run asymmetries (Nasr et al., 2018). The NARDL model adopted is described as follows: 1y t D n 1 X iD1 b1y t i C n 1 X iD0 (c C 1i 1LER C t i C c 2i 1LER t i ) C n 1 X iD0 (d C 1i 1LFR C t i C d 2i 1LFR t i )C n 1 X iD0 (e C 1i 1LPR C t i C e 2i 1LPR t i )C0 1 y t 1 C0 C 2 LER C t 1 C0 3 LER t 1 C0 C 4 LFR C t 1 C0 5 LFR t 1 C0 C 6 LPR C t 1 C0 7 LPR t 1 C u t (5) where y t represents NFPS and NFPB, which implies two NARDL equations were estimated. Table 2. Descriptive statistics. Economic risk Financial risk Political risk Mean 34.75027 38.16667 67.51167 Median 34.50000 38.50000 66.50000 Maximum 38.50000 42.00000 77.00000 Minimum 29.00000 31.50000 61.50000 SD 2.103751 1.938577 3.696337 Skewness 0.058291 0.596396 0.658080 Kurtosis 2.199080 3.064181 2.770212 Jarque–Bera 8.188301 17.83593 22.31353 Probability .016670 .000134 .000014 Sum 10,425.08 11,450.00 20,253.50 Sum of squares 363,597.6 438,132.0 1,371,433 Observations 300 300 300 4 Results discussion 4.1 Descriptive analysis Table 2 summarises the descriptive statistics for the series in question. The p values for the Jarque–Bera test for normality are all below the 5-percent level of signicance, which leads to the rejection of the null hypothesis of normality. This indicates that the coun- try risk data is not distributed normally. This result is supported by the kurtosis gures, which are below 3 for economic and political risk, suggesting a platykur- tic distribution, and above 3 for nancial risk, which suggests a leptokurtic distribution. In addition to this, the skewness values show that economic nancial risk has a negative skewness, which signals that the majority of South Africa’s economic and nancial risk ratings are clustered towards higher values, which indicate lower risk levels, whereas political risk has a positive skewness. This positive skewness indicates that the majority of the risk ratings of the country are clustered around lower values, which points to higher political risk levels. The skewness of the coun- try risk scores necessitated the use of a logarithm in the subsequent analysis. The average scores for eco- nomic and political risk fall within the moderate risk range, whereas the nancial risk score is classied as low risk. Figs. 1 and 2 illustrate NFPS and NFPB trends dur- ing the sample period from January 1995 to December 2019. In the analysis of the trendlines, a point of note is the sharp decrease in NFPS in late 2008 and early 2009, which is not reciprocated in magnitude in NFPB. This is due to the 2008 global nancial crisis, in which investors rushed to move their shareholdings to “safe haven” bond holdings. NFPS movements seem more volatile than NFBS movements and oc- cur either inversely to NFPB or drastically more than NFPB movements. 4.2 Unit root test results To assess the order of integration and station- arity of the variables used in this paper, the ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 135 Fig. 1. Net foreign purchases of shares (January 1995–December 2019). Fig. 2. Net foreign purchases of bonds (January 1995–December 2019). Kwiatkowski–Phillips–Schmidt–Shin (KPSS) and augmented Dickey–Fuller (ADF) tests in conjunction with a unit root test with breakpoints were performed. The unit root tests were carried out at constant and at constant and trend. The results of the unit root test are summarised in Table 3. The test results indicate that the variables are either I(0) or I (1). In analysing the unit root results and stationarity tests, it is prudent to note that stationarity does not change in the presence of a trend. This indicates that the trend is not signicant in the assessment of results. Interestingly, the KPSS results reect that all variables except NFPS and LER and intercept are I(1). The KPSS result for LER moves from stationarity to nonstationarity with the inclusion of a trend. This may indicate that the trend of LER is signicant. However, the ADF test and the unit root test with breakpoints do not support this. The unit root test results with breakpoints correlate with the ADF test results in that all variables besides LPR are I(0). There are no discernible patterns in the break dates, with the only similarity in break being LFR and LER in November/December 2008. This pattern represents a break during the fourth quarter of 2008. The reason for this may be linked to the nancial crisis of 2008, which caused a recession in South Africa (Verick & Islam, 2010) and would have affected the economic risk and nancial risk ratings of South Africa. 4.3 Log-run relationship analysis In this study, we identied models for log-run rela- tionship examination based on the Schwarz Bayesian information criterion. The best models were chosen for NFPS and NFBP , that is, NFPS ARDL (2, 0, 0, 0), NFPS NARDL (2, 0, 0, 0, 0, 0, 0), and NFPB (1, 0, 1, 0), NFPB (1, 0, 0, 0, 2, 0, 0) with the lag number on explanatory variables indicated in parentheses. The ARDL & NARDL F-bounds test results, displayed in Table 4, show that there is cointegration between the 136 ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 Table 3. ADF, KPSS, and breakpoint unit root test results. At level Order of integration Break dates Variable ADF KPSS Breakpoint ADF KPSS Breakpoint Breaking date NFPS At constant/intercept 6.4357* 0.8787* 10.3811* I(0) I(0) I(0) 1996M02 At trend and constant 7.0676* 0.2376* 11.3405* I(0) I(0) I(0) 1996M06 NFPB At constant/intercept 14.1696* 0.0806 14.8804* I(0) I(1) I(0) 2018M06 At trend and constant 14.1442* 0.0809 15.1441* I(0) I(1) I(0) 2018M03 LER At constant/intercept 2.5883*** 0.8026* 4.6101*** I(0) I(0) I(0) 2008M11 At trend and constant 3.2126*** 0.1707 5.2145*** I(0) I(1) I(0) 2008M11 LFR At constant/intercept 4.7854* 0.1264 5.2273* I(0) I(1) I(0) 2008M12 At trend and constant 4.8005* 0.0873 5.6724*** I(0) I(1) I(0) 2002M04 LPR At constant/intercept 2.4793 1.0841 3.7165 I(1) I(1) I(1) 1998M03 At trend & constant 2.7659 0.1139 4.8092 I(1) I(1) I(1) 2003M08 Note. *, **, and *** represent the rejection of H 0 at 1%, 5%, and 10%, respectively. The absence of any such representation indicates a failure to reject H 0 at any level of signicance. Table 4. ARDL and NARDL bounds test results. Critical values Model SBIC F-statistic Lower bound Upper bound NFPS ARDL (2, 0, 0, 0) 20.38 12.62 3.47 6.36 Long-run equation: NFPS = 3779.53LER - 15013.21LFR + 701.97LPR (6) NARDL (2, 0, 0, 0, 0, 0, 0) 20.41 8.34 2.53 4.90 Long-run equation: NFPSD 2207:09LER POS C 24212:31LER NEG 13092:74LFR POS 12476:13LFR NEG 33817:25LPR POS 12596:26LPR NEG (7) NFPB ARDL (1, 0, 1, 0) 20.72 53.65 3.47 6.36 Long-run equation: NFPBD 3167:82LER 782:14LFR18672:08LPR (8) NARDL (1, 0, 0, 0, 2, 0, 0) 20.7632 26.8098 2.53 4.9 Long-run equation: NFPBD 1956:62LER POS C 17617:05LER NEG 10167:83LFR POS 15692:39LFR NEG 19263:67LPR POS 22629:72LPR NEG (9) Note. Critical values are from Case V: unrestricted intercept and unrestricted trend. country risk measures and both NFPS and NFPB at the 5% level of signicance. This is evident in that the F statistics for NFPS (12.62) and NFPB (53.65) are greater than their respective upper and lower bound Pesaran critical values. Therefore, the null hypothe- sis of no cointegration is rejected, indicating that a long-run relationship exists between disaggregated country risk and NFPS and NFPB. Since the Schwarz Bayesian information criterion results were so close to each other, it was prudent to run an additional test to assess the symmetry of the long-run relationships. The Wald test was conducted, and its results are summarised in Table 5. The results presented in Table 5 indicate asym- metric long-run relationships, suggesting that the NARDL results are preferable to those from the ARDL model. Eqs. (6) to (9) in Table 4 represent the respective long-run relationships for NFPS and NFPB. Given the conrmation of asymmetric long- Table 5. Wald test for long-run asymmetries. Model F-statistics P-values Specication ARDL (NFPS) 17.11620 0.0000 Asymmetric NARDL (NFPS) 11.75465 0.0000 Asymmetric ARDL (NFPB) 6.165268 0.0000 Asymmetric NARDL (NFPB) 5.046736 0.0000 Asymmetric run relationships, the discussion focuses solely on NARDL Eqs. (7) and (9). Eq. (7) demonstrates that a positive change in the local economic risk (LER) is negatively correlated with NFPS, while a neg- ative change positively impacts NFPS. Specically, a 1% increase in LER is associated with a de- crease of 2,207.09 million rand in NFPS, whereas a 1% decrease in LER corresponds to an increase of 24,212.91 million rand in NFPS. Since an increase in the country risk score indicates lower economic risk, the results suggest that low economic risk lev- els decrease NFPS, while high economic risk levels ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 137 increase NFPS. This phenomenon may be attributed to higher expected returns for foreign equity in- vestors, as some developing economies with elevated risk levels are perceived as attractive, protable in- vestments (Anarkulova, 2023). Overall, this nding highlights the sensitivity of foreign investors to eco- nomic risk levels in South Africa and conrms our rst hypothesis, namely that high risk is associated with high returns. Regarding the natural log of nancial risk (LFR) and the natural log of political risk (LPR), the results indicate that both positive and negative changes in these risks have a negative long-run relationship with NFPS. This implies that lower levels of nancial and political risk decrease NFPS, while increases in these risks also result in a decline in NFPS. Although the nding that high country risk levels lead to a decrease in foreign investment ows aligns with the conclu- sions of Sekkat and Veganzones-Varoudakis (2007), the present study hints at deeper nuances in the re- lationship between nancial and political risk and NFPS. Foreign equity investors may perceive a South African market with either excessively low or high political and nancial risk levels as unattractive, re- ecting concerns about the perceived future growth potential of the country. This perception may stem from South Africa’s volatile political landscape, its status as a developing country, and a poor nancial track record (Ajide & Alimi, 2019). While both nega- tive and positive movements in LPR negatively affect NFPS, increases in political risk ratings exert a more substantial inuence on the decline of NFPS. Eq. (9) assesses the effect of disaggregated coun- try risk scores on NFPB. The equation reveals that both positive and negative changes in LER are pos- itively related to NFPB. Specically, a 1% increase in LER leads to an increase of 1,956.62 million rand in NFPB, while a 1% decrease in LER also results in a 17,617.05 million rand increase in NFPB. This suggests that foreign bond investors are not discour- aged by either low or high economic risk levels when investing in the South African bond market. This re- sult may be counterintuitive; however, it reects the complex nature of investor psychology according to behavioural nance as stated in our second hypoth- esis. Studies such as Baker and Wurgler (2007) and Ahmad et al. (2017) have shown that due to biases such as herding or familiarity bias, even institutional investors may make irrational investment decisions such as investing in a high economic risk environ- ment. Moreover, as proposed in the prospect theory by Kahneman and Tversky (1979), the way in which risk is perceived by investors frames their investment decisions, which may be contrary to rational thought. This could indicate that there is a certain level of high economic risk that makes bond investors attracted to bond investment with expectation of high returns as portrayed by the risk-return relationship. Conversely, both positive and negative changes in LFR and LPR show a negative association with NFPB. The results indicate that a 1% increase in LFR and LPR decreases NFPB by 10,167.83 million rand and 19,263.67 million rand, respectively. This nding indicates that a South Africa characterised by high nancial and political risk leads to outows of foreign bond investments from the South African market. Unstable exchange rates, limited international liquidity, and rising cor- ruption can all contribute to elevated nancial and political risk. Given that bond investors are partic- ularly sensitive to interest rates—factors that can be negatively impacted by these conditions—they may be discouraged from investing in the country (Fabozzi & Fabozzi, 2021). Contrary to expectations, a 1% de- crease in LFR and LPR corresponds to decreases in NFPB of 15,692.39 million rand and 22,629.72 million rand, respectively; a relationship similar to that of NFPS and these exact country risk measures. Such a nding is in line with our second hypothesis as it implies that political and nancial risk have a more nuanced dynamic with NFPS and NFPB. Declining political and nancial risk levels may be associated with lower returns, making the South African market less attractive for investors. These results then sug- gest that there could exist an optimal political and nancial risk range between which foreign equity and bond investors are not disincentivised by excessive political and nancial instability and excessive politi- cal and nancial stability. These ndings also conrm our third prior expectation, namely that the effect of country risk on net foreign portfolio investments may differ across the components of country risk. A closer examination of the results reveals that negative changes in local political risk (LPR) have the strongest effect on NFPB, which aligns with the ndings of Muzindutsi and Obalade (2024), who identied political risk as a driver of bond returns. During the observation period, South Africa’s po- litical environment has undergone signicant trans- formation, including the establishment of a new democracy, a presidency marked by corruption under Jacob Zuma, and the “Fees Must Fall” movement, which highlighted the levels of inequality still faced in the country post-apartheid. These events, among others, contribute to political uncertainty, adversely affecting foreign investor condence. Therefore, this nding emphasises the importance of policymak- ers strengthening governance and political stability while enhancing security and social stability. The ndings of this study are consistent with those of Baek and Qian (2011), and Meyer and Habanabakize (2018). 138 ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 Table 6. Error correction models. Model Variables Coefcients Probability NFPS (ARDL) C 14,880.99 0.0000 @TREND 13.67202 0.0025 D(NFPS( 1)) 0.269752 0.0000 CointEq( 1) 0.445266 0.0000 NFPS (NARDL) C 876.1663 0.2178 @TREND 93.22322 0.0000 D(NFPS( 1)) 0.235367 0.0001 CointEq( -1) 0.514832 0.0000 NFPB (ARDL) C 58,748.53 0.0000 @TREND 5.030486 0.2998 D(LFR) 64,366.98 0.0000 CointEq(-1) 0.826954 0.0000 NFPB (NARDL) C 650.3266 0.4600 @TREND 8.572711 0.0771 D(LFR_NEG) 109,172.3 0.0000 D(LFR_NEG( 1)) 86,029.15 0.0005 CointEq( 1) 0.790281 0.0000 Table 7. Diagnostic tests. Serial correlation test White heteroscedasticity test Model (F stat. p values) (F stat. p values) CUSUM test NFPS ARDL 0.066020 4.013455 Stable NFPS NARDL 0.219669 3.508220 Stable NFPB ARDL 4.458889 3.630707 Stable NFPB NARDL 3.325175 1.769112 Stable 4.4 Short-run relationship analysis Following the establishment of long-run relation- ships between disaggregated country risk and FPI ows, ECMs were estimated for both ARDL and NARDL equations of NFPS and NFPB. This was done to further analyse the short-run dynamics between the variables along with adjustment to the long-run equilibrium, and the results are summarised in Ta- ble 6. In reviewing the error correction term (ECT) in all four ECMs, it is notable that all are negative and statistically signicant, which is the desired result for a short-run relationship to be valid. The ARDL ECTs for the NFPS and NFPB show that 44.53% and 82.70% of any disequilibrium are corrected monthly, respectively. On the other hand, the NARDL ECTs for NFPS and NFPB indicate that about 51.48% and 79.03% of any deviation from equilibrium are cor- rected monthly, respectively. The higher ECTs for NFPB implies that foreign bond investments adjust back to equilibrium at a faster rate than foreign eq- uity investment ows. The results further reveal that only negative changes in LFR have a statistically sig- nicant impact on NFPB in the short run. Financial crises, which are commonly associated with growing nancial risk, may hamper the government’s ability to honour the bondholders’ high-return payments. It is therefore crucial that policymakers acknowledge the short-term relation between nancial risk and FPI so as to effectively manage nancial risk in a manner that avoids nancial crises and encourages the foreign purchasing of bonds in South Africa. 4.5 Diagnostic tests The diagnostic tests of the estimated models are indicated in this subsection. Table 7 summaries the results obtained for the Breusch–Godfrey serial cor- relation tests and the white heteroscedasticity test. The results show that the residuals from the models in question are all not serially correlated and are ho- moscedastic at a 5% level of signicance. The CUSUM (cumulative sum) test for stability shows stable mod- els as the graph did not cross the 5% boundaries during the sample period. The overall outcome of the results proves that this study does not contravene any economic inferences and justies the credibility of the outcomes in this assessment. 5 Conclusions Through the application of ARDL and NARDL models, this study investigates the short-run and long-run effects of disaggregated country risk on FPI ows in South Africa. This study analysed the im- pacts of various elements of country risk, namely economic, nancial, and political risk, on FPI ows. Our ndings reveal a cointegrating relationship ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 139 between disaggregated country risk and FPI ows, as measured by NFPS and NFPB. The effects of in- dividual country risk measures on the two types of FPI ows were found to be asymmetric. Specically, low levels of economic risk are associated with a de- cline in foreign equity ows but an increase in foreign bond investments in the long run. Conversely, high economic risk levels in South Africa are linked to ris- ing foreign equity and bond investment ows. This suggests that foreign investors may view South Africa as an attractive investment opportunity during times of high economic risk due to the potential for high returns. In terms of nancial and political risk, both high and low levels of these risks were linked to a decrease in net foreign equity and bond investment ows. While foreign investors may be willing to invest in a highly uncertain economic climate, the same cannot be said for circumstances characterised by escalating nancial and political risks. This nding revealed the existence of a possible optimal political and nan- cial risk threshold where equity and bond returns are maximised. Moreover, low levels of political risk had the most signicant impact on foreign equity ows, while rising political risk levels had the greatest effect on foreign bond investments in the long run. This nding conrms our hypothesis that the effect of eco- nomic, nancial, and political components of country risk on net foreign portfolio ows varies because in- vestors may react differently to each of these risk components. In this case, the political risk is found to be a prominent determinant of the shift in net portfolio investment inows among these risks. This process underscores the importance of policymakers stabilising political risk to encourage FPI. In the short run, nancial risk was the only country risk measure that signicantly affected NFPB. The nature of the re- lationships between disaggregated country risk mea- sures and NFPS and NFPB highlighted by this study emphasises the need for policymakers to understand these intricate dynamics. Such understanding is cru- cial for implementing policies that foster a mutually benecial economic, political, and nancial environ- ment in South Africa, thereby encouraging FPI while maintaining the country’s sovereignty. Although this study provides valuable insights for policymakers and scholars, the observation period is limited to 2019 due to data constraints. Future research could explore these relationships while accounting for recent eco- nomic, political, and nancial developments in South Africa. Given that political risk had the most signif- icant impact on both NFPS and NFPB, it would be worthwhile for future studies to closely investigate which individual political risk components most de- ter foreign investment ows into South Africa. References Ahmad, Z., Ibrahim, H., & Tuyon, J. (2017). Institutional investor behavioral biases: Syntheses of theory and evidence. Manage- ment Research Review, 40(5), 578–603. https://doi.org/10.1108/ MRR-04-2016-0091 Ajide, K. B., & Alimi, O. Y. (2019). Political instability and migrants’ remittances into sub-Saharan Africa region. GeoJournal, 84(6), 1657–1675. https://doi.org/10.1007/s10708-018-9942-8 Allen, D. E., & McAleer, M. (2021). A nonlinear autoregressive dis- tributed lag (NARDL) analysis of the FTSE and S&P500 Indexes. Risks, 9(11), Article 195. https://doi.org/10.3390/risks9110195 Al Samman, A., & GabAlla, M. K. (2020). Impact of country risk and return on FPI. International Journal of Economics and Financial Issues, 10(6), 57–68. https://doi.org/10.32479/ije.10495 Anarkulova, A. (2023). The risk-return tradeoff: Evidence from a broad sample of developed markets [Doctoral dissertation, Emory Univer- sity]. Baek, K., & Qian, X. (2011). An analysis on political risks and the ow of foreign direct investment in developing and industrial- ized economies. Economics, Management, and Financial Markets, 6(4), 60–91. Bah, S. I., & Giritli, N. (2020). What drives foreign portfolio invest- ment ows in South Africa? Journal of Yasar University, 15(58), 368–380. https://doi.org/10.19168/jyasar.598067 Baker, M., & Wurgler, J. (2007). Investor sentiment in the stock market. Journal of Economic Perspectives, 21(2), 129–151. https:// doi.org/10.1257/jep.21.2.129 Benoît, J. P ., & Dubra, J. (2011). Apparent overcondence. Economet- rica, 79(5), 1591–1625. https://doi.org/10.3982/ECTA8583 Chorn, S., & Siek, D. (2017). The impact of foreign capital inow on economic growth in developing countries. Journal of Finance and Economics, 5(3), 128–135. Elton, E. J., Gruber, M. J., Brown, S. J., & Goetzmann, W. N. (2009). Modern portfolio theory and investment analysis (8th ed.). John Wiley & Sons. Fabozzi, F. J., & Fabozzi, F. A. (2021). Bond markets, analysis, and strategies (10th ed.). MIT Press. Haider, M. A., Khan, M. A., Saddique, S., & Hashmi, S. H. (2017). The impact of stock market performance on foreign portfolio investment in China. International Journal of Economics and Fi- nancial Issues, 7(2), 460–468. Hassan, A. S. (2023). Does country risk inuence foreign direct investment inows? Evidence from Nigeria and South Africa. Journal of Contemporary Management, 20(1), 658–679. https:// doi.org/10.35683/jcm22022.212 Havemann, R., van Vuuren, H. J., Steenkamp, D., & van Jaarsveld, R. (2022). The bond market impact of the South African Reserve Bank bond purchase programme. South African Reserve Bank. https://www.resbank.co.za/content/dam/ sarb/publications/working-papers/2022/wp%202203.pdf Hayakawa, K., Kimura, F., & Lee, H. H. (2013). How does coun- try risk matter for foreign direct investment? The Developing Economies, 51(1), 60–78. https://doi.org/10.1111/deve.12002 Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291. https://doi .org/10.2307/1914185 Kartal, M. T., Ertu˘ grul, H. M., & Ulussever, T. (2022). The impacts of foreign portfolio ows and monetary policy responses on stock markets by considering COVID-19 pandemic: Evidence from Turkey. Borsa Istanbul Review, 22(1), 12–19. https://doi.org/10 .1016/j.bir.2021.06.003 Khan, M. M., & Akbar, M. I. (2013). The impact of political risk on for- eign direct investment (MPRAPaper No. 47283). Munich Personal RePEc Archive. https://mpra.ub.uni-muenchen.de/47283/ Magoane, R. M., Meyer, D. F., & Muzindutsi, P . F. (2023). The asym- metric effect of political risk and exchange rate uctuations on foreign direct investment inows in South Africa. The Journal of Developing Areas, 57(2), 253–268. https://doi.org/10.1353/jda .2023.0032 Malala, O. N., & Adachi, T. (2020). Portfolio optimization of electric- ity generating resources in Kenya. The Electricity Journal, 33(4), Article 106733. https://doi.org/10.1016/j.tej.2020.106733 140 ECONOMIC AND BUSINESS REVIEW 2025;27:130–140 Mamvura, K., & Sibanda, M. (2020). Modelling short-run and long-run predictors of foreign portfolio investment volatility in low-income Southern African Development Community coun- tries. Journal of Economic and Financial Sciences, 13(1), Article 559. https://doi.org/10.4102/jef.v13i1.559 McCue, C. P . (2000). The risk-return paradox in local government investing. Public Budgeting & Finance, 20(3), 80–101. https://doi .org/10.1111/0275-1100.00021 Meyer, D. F., & Habanabakize, T. (2018). An analysis of the relation- ship between foreign direct investment (FDI), political risk and economic growth in South Africa. Business and Economic Hori- zons, 14(4), 777–788. https://doi.org/10.22004/ag.econ.287229 Mittal, S. K. (2022). Behavior biases and investment decision: Theoretical and research framework. Qualitative Research in Fi- nancial Markets, 14(2), 213–228. https://doi.org/10.1108/QRFM -09-2017-0085 Muguto, H. T., Rupande, L., & Muzindutsi, P . F. (2019). Investor sentiment and foreign nancial ows: Evidence from South Africa. Zbornik radova Ekonomskog fakulteta u Rijeci, 37(2), 473– 498. https://doi.org/10.18045/zbefri.2019.2.473 Mutize, M., & Gossel, S. J. (2019). Sovereign credit rating announce- ment effects on foreign currency denominated bond and equity markets in Africa. Journal of African Business, 20(1), 135–152. https://doi.org/10.1080/15228916.2019.1580996 Muzindutsi, P . F., & Mjeso, T. (2018). Analysis of South African household consumption expenditure and its determinants: Ap- plication of the ARDL model. EuroEconomica, 37(3), 169–179. Muzindutsi, P . F., & Obalade, A. A. (2024). Effects of country risk shocks on the South African bond market performance un- der changing regimes. Global Business Review, 25(1), 137–149. https://doi.org/10.1177/0972150920951116 Muzindutsi, P . F., Rajhununan, S., Dube, M., Ganie, B., Mahess, E., & Reddy, D. (2022). The effect of economic, nancial and political country risk factors on the JSE mining index: An ARDL approach. International Journal of Trade and Global Markets, 15(1), 22–31. https://doi.org/10.1504/IJTGM.2022.120905 Nasr, A. B., Cunado, J., Demirer, R., & Gupta, R. (2018). Country risk ratings and stock market returns in Brazil, Russia, India, and China (BRICS) countries: A nonlinear dynamic approach. Risks, Measuring and Modelling Financial Risk and Derivatives, 6(3), 94– 116. https://doi.org/10.3390/risks6030094 National Treasury. (2022). Budget review 2022. http://www. treasury.gov.za/documents/national%20budget/2022/review/ Prelims.pdf Nhlapho, R., & Muzindutsi, P . F. (2020). The impact of disag- gregated country risk on the South African equity and bond market. International Journal of Economics and Finance Studies, 12(1), 189–203. https://doi.org/10.34109/ijefs.202012112 OECD (2018). Main economic indicators, volume 2018, issue 2. OECD Publishing. https://doi.org/10.1787/mei-v2018-2-en Oleksiv, M. (2000). The determinants of FDI: Can tax holiday com- pensate for weak fundamentals? Case of Ukraine [Master’s thesis, National University of Kyiv-Mohyla Academy]. The PRS Group. (2022). The international country risk guide (ICRG). https://www.prsgroup.com/explore-our-products/icrg/ Rafat, M., & Farahani, M. (2019). The country risks and foreign direct investment (FDI). Iranian Economic Review, 23(1), 235–260. https://doi.org/10.22059/ier.2018.69107 Schwab, K. (2017). The Global Competitiveness Report 2017–2018. The World Economic Forum. https://www.weforum.org/ publications/the-global-competitiveness-report-2017-2018/ Sekkat, K., & Veganzones-Varoudakis, M. A. (2007). Openness, investment climate, and FDI in developing countries. Review of Development Economics, 11(4), 607–620. https://doi.org/10 .1111/j.1467-9361.2007.00426.x Shin, Y., Yu, B., & Greenwood-Nimmo, M. (2011). Modelling asym- metric cointegration and dynamic multipliers in an ARDL framework. In W. C. Horrace & R. C. Sickles (Eds.), Festschrift in honor of Peter Schmidt (pp. 281–314). Springer Science & Business Media. https://doi.org/10.1007/978-1-4899-8008-3_9 Singhania, M., & Saini, N. (2018). Determinants of FPI in developed and developing countries. Global Business Review, 19(1), 187–213. https://doi.org/10.1177/0972150917713280 Topal, M. H., & Gül, Ö. S. (2016). The effect of country risk on foreign direct investment: A dynamic panel data analy- sis for developing countries. Journal of Economics Library, 3(1), 141–155. van Wyk, J., & Lal, A. K. (2008). Risk and FDI ows to develop- ing countries. South African Journal of Economic and Management Sciences, 11(4), 511–528. https://doi.org/10.4102/sajems.v11i4. 285 Verick, S., & Islam, I. (2010). The great recession of 2008-2009: Causes, consequences and policy responses (IZA Discussion Paper No. 4934). Institute for the Study of Labor. World Bank Group. (2018). South Africa—Systematic country diagnos- tic: An incomplete transition: Overcoming the legacy of exclusion in South Africa. https://documents.worldbank.org/curated/en/ 815401525706928690 Zaimovic, A., Omanovic, A., & Arnaut-Berilo, A. (2021). How many stocks are sufcient for equity portfolio diversication? A re- view of the literature. Journal of Risk and Financial Management, 14(11), Article 551. https://doi.org/10.3390/jrfm14110551