Exchange Rates, Supply Chain Activity/Disruption Effects, and Exports

Abstract

In the past, South African monetary policy aimed to protect the external value of the domestic currency (Rand); however, these efforts failed. Later, its monetary policy approach changed to allow the foreign exchange rate market to determine the exchange rates. In such a change, the South African Reserve Bank (SARB) aimed to stabilize the demand for the Rand in the foreign exchange market by providing information to stabilize market expectations and create favorable market conditions. However, South African policymakers have struggled with currency depreciation since the early 60s, increasing the uncertainty of South African exports. This study aims to examine the effect of currency depreciation on exports using the Threshold Autoregressive (TAR) model. Additionally, this study created and validated the supply chain activity/disruption index to capture the sea trade activity. The sample period for the analysis is 2009 to 2023. The study finds that currency depreciation does not improve trade between South Africa and its trading partners over time. Furthermore, the currency depreciation was found to be asymmetric to the effect of international trade across the different regimes. The supply chain activity index shows that the effect of supply chain activity/disruption on exports is regime-dependent. This implies that the effect on exports is dependent on the economic environment.

1. Introduction

The Commission on Growth and Development in 2008 seems to have inspired the government’s substantial reliance on the belief that export promotion will drive economic growth. The commission pointed out that the government should invest more effort in export growth. As stated, the South African government’s overall strategy includes export promotion as a driving force for creating jobs and boosting investments. The South African government seeks to achieve this by increasing domestic goods’ competitiveness. In 2010, the then director general of the National Treasury of South Africa, Lesetja Kganyago, supported the assertion that the uncompetitiveness of the Rand contributes to low export growth [1]. Furthermore, in 2008, the Rand depreciated against the United States of American Dollar (USD), and the exports increased. However, Kganyago notes that this improvement is at the expense of higher domestic inflation [1].

Another point to consider is the National Development Plan (NDP), introduced by the South African government in 2012 as a macroeconomic strategy for developing the economy. In the NDP, it is clear that growing exports are a central driver of economic growth. Furthermore, it highlights the necessity of aiding small and medium enterprises to enable them to access the international market through exports. Among many scholars, ref. [2] detail the deliberate effort of the South African government to export more.

On the other side, the monetary authority of South Africa, the South African Reserve Bank (SARB), has independence from the government. The central bank’s autonomy is derived from the constitution of the South African Republic. As such, the central bank’s primary objective is to achieve the internal price stability of the Rand. Hence, the SARB targets the inflation rate range of 3–6%, which is why the bank is committed to keeping the inflation rate within the range [3]. Therefore, any force that could potentially put pressure on the domestic price level to exceed the band on the upper limit is met with a hawkish response by the SARB [4].

The SARB, through the Monetary Policy Committee (MPC), makes announcements, which take place every quarter during the year, always giving a warning that it will not hesitate to counteract any force likely to lead to an inflation rate overshooting the band. The ref. [3], in the monetary policy review released in April 2024, provides a precise analysis of how the depreciation of the exchange rate leads to an increase in the inflation rate. However, the SARB does not intervene in the foreign exchange market, but it does pay attention to the exchange rate dynamics. Therefore, the central bank acts without hesitation when there is pressure on the domestic price level, which is by increasing the interest rate. This is also supported by [5,6], who note that the SARB does not view the depreciation of the Rand as favorable.

Despite the SARB managing to contain the inflation rate within the targeted range to achieve its objective, export promotion may be perceived as being sacrificed by the central bank when it chooses to protect the internal value of the Rand. This issue of policy contradiction has grown in importance in recent years because of the low economic growth South Africa has experienced in the last two decades. Hence, there is continuous debate about the best approach for the management of the economy in order to boost exports, which the government views as an organic antidote to the economic ills facing South Africa.

Studies on this subject have not included the effect of supply chain disturbances as they have implications on exports and the price level of imported goods [7,8]. One of the key components in the supply chain mix is the transportation of goods by sea, which is vital for exports; this was evident during the COVID-19 pandemic. The pandemic led to a limit of goods transported by sea, leading to shortages of goods and services in all parts of the world since exports were restricted. Post-COVID-19, there have been many disruptions to the transportation of goods and services around the world, introduced by conflicts that have led to the prevention of shipping. Most studies on this subject have focused on monetary policy effectiveness following a supply chain disruption, such as [9], who argue that monetary policy is effective in halting the effects of supply chain disruptions. Hence, much of the literature links some components of monetary policy, such as exchange rate depreciation and international trade [10,11]. However, far too little attention has been given to considering the effects of supply chain activity on countries’ export ambitions. Notably, there is inadequate quantitative analysis linking supply chain activities and exports due to the unavailability of measures for supply chain activity/disruptions. As such, part of the objective of this study is to develop an index that measures supply chain activities/disruptions. Therefore, the study makes a scientific contribution by providing a tested and validated method to capture these supply chain activities.

The remainder of the manuscript is as follows: exchange rate depreciation fundamentals, methodology, data and variables, results, and conclusion.

2. Literature Review

Studies have examined the link between currency fluctuation and exports. Ref. [12] explored the impact of currency depreciation on exports in selected Asian countries to test whether depreciation would have a positive effect on the trade balance. However, the results of the study showed that depreciation had no significant effect on the trade balance. The authors suggested this could be due to a decline in primary commodities and manufactured products’ trade or heavy reliance on imported goods, negatively affecting the local currency. According to [13], exchange rate fluctuations have an asymmetric effect on the trade balance. This means that even if the degree of currency depreciation is the same, its impact on the trade balance can vary. The studies noted that changes in currency can affect the trade balance because they can affect the prices of goods traded.

Ref. [10] illustrated that a real exchange rate depreciation causes a decline in imports and an increase in exports. In addition, the authors observed that imports impact exports, implying that a decrease in imports relating to a real exchange rate depreciation will negatively affect exports. They concluded that a real exchange rate depreciation does not improve the trade balance. This view is supported by the other literature, such as studies by [10,14,15,16], which also confirm that currency depreciation worsens the trade deficit. This is primarily due to higher import prices greater than the reduced volume. Ref. [6] conducted a study on the impact of exchange rates on the trade balance and output growth, specifically focusing on the asymmetric effects of the exchange rate. The authors highlighted that several African countries have experienced economic recession due to a decline in their foreign national reserve and high import bills due to the exchange rate threat to Africa’s trade balance.

Ref. [17] also emphasized that African nations have experienced significant exchange rate movements since the global crisis of 2008/9. Ref. [6] further noted that highly unstable and uncertain exchange rates could hinder the performance of macroeconomic indicators such as price and output, total exports, and external competitiveness. The authors concluded that currency fluctuation is crucial in regulating trade direction in South Africa. Ref. [18] conducted a study to understand the impact of currency depreciation on the Nigerian economy. The study revealed that due to the implementation of economic reforms, the value of the Nigerian currency has undergone significant changes, such as depreciation. Different exchange rate policies were implemented to make the exchange rate market-driven, which resulted in continuous depreciation of the Nigerian Naira against major international currencies. The results showed that the currency’s depreciation positively impacts the domestic output level in the long run but has a negative effect in the short run. Therefore, a decrease in the local currency leads to fluctuations in domestic output and prices. Overall, the study concluded that currency depreciation significantly impacts the Nigerian economy and suggested that policymakers implement measures to stabilize the currency to avoid fluctuations in domestic output and prices.

Ref. [19] investigated the effectiveness of buyer–supplier currency exchange rate flexibility contracts in global supply chains. They examined two types of currency exchange rate flexibility to determine the characteristics of exchange rate risk mitigation policies for both buyers and suppliers. The results indicated that when the wholesaler price is uncertain due to exchange rate volatility, the optimal order quantity of the buyer decreases. Furthermore, the proposed contracts appear to be advantageous for both buyer and supplier when payment is made in the supplier’s currency, indicating a preference for implementing such contractual agreements from the perspective of both parties. Conversely, when payment is made in the buyer’s currency, the proposed contracts do not result in a mutually beneficial situation for both parties. Ref. [20] examined the connection between oil and global foreign exchange markets, specifically the role of economic policy uncertainty. The authors investigated the volatility spill-over between crude oil and exchange rates and found a strong correlation between them, with oil being a net receiver of the shock. Additionally, they conducted a nonparametric quantities-based causality test and demonstrated that the spillover for each asset is motivated by economic policy uncertainty around lower and median quantiles. This implies that the role of the U.S. economic policy in influencing global financial cycles, leading to capital flows and price movements in asset prices, cannot be overemphasized. Ref. [21] conducted a study on the relationship between the news, sentiments, and capital flows. The authors investigated the impact of two types of shocks on gross capital flows expectation news (which refers to increases in expected future productivity) and sentiments (which represent surges in optimism that are not associated with future productivity). The study found that both shocks account for over 80% of the variation in gross capital flows across all time horizons, with sentiment shocks having the greatest impact. While both shocks show a positive correlation between gross inflows and outflows, only sentiment shocks generate procyclical gross flows.

3. Methodology

This study adopts the imperfect substitute model. The model has enjoyed enormous empirical validation [10,22]. It assumes imports and exports are not perfect substitutes for consumption of nontraded goods produced domestically. There is extensive empirical support for this assumption, among others [23]. Furthermore, assume a small open economy where time is continuous for a representative utility-maximization household. Following [10], we assume households are identical and have acquired an internationally traded debt accruing from previous transactions. An additional assumption of full capital mobility is imposed on the model to make this assumption valid. Therefore, the representative household takes the interest rates as given. So, this makes the exchange rate depreciation vary in the prices of nontraded and traded goods. Thus, these assumptions are illustrated in different scenarios.

3.1. Modeling Demand for Exports by a Foreign Country

The representative household consumes nontraded and traded goods. For this model, the imported goods are assumed to be produced by the big economy. On the other side, the home country has an endowment of goods. Households in foreign countries are assumed to be lenders, and they earn interest on income. They choose between accumulating assets or consumptions. Therefore, the representative consumer’s problem is given by

$$\max U={{\displaystyle \int }}_{t=0}^{\infty }\left[\phi ln\left(l_t\right)+\left(1-\phi \right)ln\left(x_t\right)\right]exp\left(-{\gamma }_t\right)$$

(1)

where $l_t$ denotes nontraded and $x_t$ represents exports. Furthermore, $\gamma$ represents the individual’s rate of time preference. The total endowment of goods is given by

$$z_t=s_t+x_t{\left(p^x/p\right)}_t\left[1+{\omega }_t\right]$$

(2)

where $s_t$ is the endowment of domestic goods while ${\left(p^x/p\right)}_t$ is the domestic price of exports goods relative to the home goods. Notably, Equation (2) is augmented from what it is in [19]. Furthermore, the budget constraint is given by

$$\dot{d}=s_t^{*}+x_t{\left(p^m/p^{*}\right)}_t-r_t^{*}d{\left({\displaystyle \frac{p^x}{p^{*}}}\right)}_t-l_t^{*}-x_t{\left({\displaystyle \frac{p^x}{p^{*}}}\right)}_t$$

(3)

where $r_t^{*}$ is the world interest rate and the relative price of imports and domestic price defined by ${\left(p^m/p^{*}\right)}_t$ and ${\left(p^x/p^{*}\right)}_t$ describes the price of the imported to the home currency. Following this description, assume the law of one price such that $m_t{\left(p^m/p^{*}\right)}_t\left[1+{\omega }_t\right]$ and the Purchasing Power Parity (PPP). To apply the law of one price, the price differences reflected in differences in distribution cost means the ordering of the price between the two countries is similar to the same goods. Therefore, the expression for the total endowment can be expressed in terms of the home good such that

$$z_t^{*}=s_t^{*}+m_t{\left(p^m/p^{*}\right)}_t\left[1+{\omega }_t\right]$$

(4)

where ${\omega }_t$ is the cost of shipping import. The cost of shipping goods is associated with supply chain disruptions. For example, if there is a blockage in the shipping route, the days of delivering the goods are going to be longer because the ships will take alternative routes. Hence, this is also reflected in shipping prices (fright rates).

The first order condition is taken between consumption and domestic and imported goods.

$$l_t^{*}=\left[\beta /\left(1-\beta \right)\right]x_t{\left(p^x/p\right)}_t$$

(5)

The variations in the model are described by the Euler equation such that

$${\dot{x}}_t=x_t\left(r_t^{*}-\alpha \right)$$

(6)

Note that the preference for time is assumed to be the same for both a developing country and a developed one.

3.2. Modeling Emerging Economies’ Import Demand

The representative household maximizes the lifetime utility function, which is

$$\max U={{\displaystyle \int }}_{t=0}^{\infty }\left[\phi ln\left(l_t\right)+\left(1-\phi \right)ln\left(m_t\right)\right]exp\left(-{\gamma }_t\right)$$

(7)

Assume that the representative household derives the utility function following the Cobb-Douglas utility function. The budget constraint is given by

$$\dot{d}=s_t+x_t{\left(p^x/p\right)}_t-r_{t}d{\left({\displaystyle \frac{p^x}{p}}\right)}_t-l_t-m_t{\left({\displaystyle \frac{p^m}{p}}\right)}_t$$

(8)

After setting up the Lagrange function, take the first order conditions and then find the solution for $l_t$ such that

$$l_t=\left[\beta /\left(1-\beta \right)\right]m_t{\left(p^m/p\right)}_t$$

(9)

Equation $\left(9\right)$ represents the intertemporal marginal rate of substitution between nontradable and importables to the relative price. Then, variations in consumption of imported goods are given by the Euler equation such that

$${\dot{m}}_t=m_t\left(r_t^{*}-\alpha \right)$$

(10)

Equation $\left(10\right)$ is the marginal rate of substitution between the current and future consumption for the relevant world interest rate.

3.3. Modeling the Equilibrium

The Euler Equations (6) and $\left(10\right)$. However, this study aims to analyze the long-run relationship where the subjective rate of time preference should be the same as the world interest rate $\left(\alpha =r^{*}\right)$. The equilibrium solution is obtained by combining the Equations (3) and $\left(10\right)$, i.e., the solution combination of import to their price relative to the domestic goods to permanent income. Therefore, the equilibrium combinations are given by

$$m_t=\left[x_t-r^{*}d\right]{\left({\displaystyle \frac{p^x}{p}}\right)}_t-{\left({\displaystyle \frac{p^m}{p}}\right)}_t$$

(11)

$$x_t={\displaystyle \frac{\left[m_t{p}_t^m+r^{*}d{p}_t^x\right]}{p^{*}}}-{\left({\displaystyle \frac{p^x}{p^{*}}}\right)}_t$$

(12)

Then, by substituting Equation (4) into (12), we obtain

$$x_t=\left(z_t^{*}-s_t^{*}\right)+\left(d{p}_t^x-1\right) {\left({\displaystyle \frac{p^x}{p^{*}}}\right)}_t-{\omega }_t$$

(13)

In Equation (13), $\left(z_t^{*}-s_t^{*}\right)$ is the global economic activity, $\left(d{p}_t^x-1\right) {\left({\displaystyle \frac{p^x}{p^{*}}}\right)}_t$ is the real effective exchange rate, and $\omega$ is the supply chain disturbance/activity.

Therefore, Equation (13) implies that exports are a function of global economic activity, exchange rate, and supply chain disruptions. This equation is estimated using the Threshold Autoregressive (TAR) model to determine the relationship between exchange rate depreciation and international trade. The TAR is an autoregression model that permits a single or more regime to be created for a threshold variable. Hence, it allows for the inclusion of asymmetry behavior that cannot be explained by other autoregressive models. The regimes can be defined as follows:

$$h_t=\left\{\begin{array}{c}{\sigma }_{11}{h}_{t-1}+\cdots +{\sigma }_{1n}{h}_{t-n}+{\epsilon }_t if k_{t-d}\le j \\ {\sigma }_{22}{h}_{t-1}+\cdots +{\sigma }_{2i}{h}_{t-i}+{\epsilon }_t if k_{t-d}>j\end{array}\right.$$

(14)

The $n$ in $\left(14\right)$ was chosen based on the information that coincides with the threshold value $k_t$. Then, $d$ is delay. In this case, $k$ is given by $\Delta h$. Therefore, the delay is a discrete parameter, estimated by searching over possible values. Hence, the values of $j$ are less discrete because they depend on $k_{t-d}$.

4. Data and Variables

The data used in the study were collected from the Federal Reserve Bank of St Louis, the South African Reserve Bank, and Bloomberg. The sample data are from the second quarter of 2009 to the last quarter of 2023. The starting date was selected because data used to construct the supply chain activity variables was unavailable before 2009. The variables used in the study are exports, supply chain activity, global economic activity, and exchange rate depreciation.

The single variable measuring supply chain disruptions/activity was constructed using the principal component analysis. In constructing the variable, economic policy uncertainty, global price of commodities, and shipping freight rates. The reason for including economic policy uncertainty is that uncertainty is an important factor in macroeconomic decisions. Ref. [24] also confirm that firms may postpone their investment decisions due to economic policy uncertainty. Economic policy uncertainty leads firms to delay their action until enough information is available. Therefore, this has an impact on supply or demand for goods. This is also confirmed by [8], who argued in favor of economic policy uncertainty as being the main factor influencing the shipping of goods and services.

Furthermore, the demand for commodities is the main driver of shipping globally. This is supported by [25], who show that when economies grow, they demand raw materials used in production. Furthermore, a sluggish demand for commodities lead to a fall in maritime activities. Ref. [26] link international trade to maritime activities and argue that it constitutes 90% of economic activities. Therefore, maritime activities will influence exchange rates because they facilitate the transportation of goods. If the goods cannot be transported, it affects international trade, which impacts the exchange rates through the current account balance [27].

Another point to consider is the shipping freight rates. When there is a blockage in the shipping routes, the ships are required to take alternative routes, which usually take longer. Hence, the freight rate would tend to be higher. The longer the days the ships spend in international waters, the more expensive the shipping costs [26]. Any disruption to the supply chain is reflected in the prices of goods transported in international waters [28]. Therefore, optimizing freight rates leads to efficient logistics (transportation of goods) [29]. Hence, the supply chain disruption index created in this study captures the transportation of goods better than available indices.

The data from the above mentioned three is used to extract a common component in order to create a variable that captures the supply chain activities. The principal component analysis is used to create an index for this variable. The following table shows the results of the principal component estimation using the data from the three variables discussed above.

The highest proportion of the first principal component can explain 45% of the variations (

Table 1. Principal Components Estimation.

				Cumulative	Cumulative
Number	Value	Difference	Proportion	Value	Proportion
1	1.345915	0.181691	0.4486	1.345915	0.4486
2	1.164224	0.674362	0.3881	2.510138	0.8367
3	0.489862	---	0.1633	3.000000	1.0000

The column with a difference is generated by subtracting the new value from the previous value. In the case of row 3 not having a value in the corresponding column, it is because there are only 3 rows. Source Authors’ estimation.

). This is a fairly good approximation of the variable. This shows no parallels, meaning it will capture the desired effects. Furthermore, in the following table, the results of the correlations between the variables are presented. In the following table, PC means principal component.

The three generated PCs have good correlations with the data of the variables (

Table 2. Correlation between variables.

Variable	PC 1	PC 2	PC 3
Economic policy uncertainty	0.769952	0.062655	−0.635018
Global prices of commodities	0.515684	−0.647222	0.561403
Freight rates	0.375823	0.759722	0.530640

Source: Authors’ estimation.

). The PC 1 has correlations of 0.77, 0.52, and 0.38, respectively, with economic policy uncertainty, global prices of commodities, and freight rates (Table 2). Whereas, the PC 2 correlation with economic policy uncertainty, global prices of commodities, and freight rates, are 0.06, −0.65, and 0.76, respectively (Table 2). Lastly, economic policy uncertainty, global prices of commodities, and freight rates are correlated with PC 3, respectively, by 0.64, 0.56, and 0.53. The first principal component fits all the variables better than the other two principal components and hence it was adopted.

Following these results, the next figure shows the distribution of the data used in the construction of the principal component. The diagram shows the biplot, which is used to visualize and interpret the scores at the same time. In the following diagram, FR denotes fright rate, ECONOMIC_POLICY_UNCERTAINTY denote economic policy uncertainty, and GLOBAL_PRICES_OF_COMMODITIES represent global prices of commodities.

In the biplot, the vectors measuring freight rates and global prices of commodity prices lie close to the origin, while economic policy uncertainty is close to the origin (Figure 1). This means both freight rates and the global prices of commodities have similar responses. Hence, they are likely to be more correlated with each other. However, if the correlation is higher than 0.5, it is likely that there is an effect of outliers. On the other hand, the vector measuring economic policy uncertainty lies close to the average values of the sample (Figure 1).

Then, to obtain a clear picture of the principal components, subsequently, an assessment of the scree plot is required. This provides more information about the components. The assessment of the scree plot assumes that the relevant information of the component is more than the random innovation contained in the component. The line of a scree plot is broken. This is estimated for random data with an expectation that the eigenvalues follow this random structure.

The eigenvalues for the first component are above one (Figure 2). This means component one is able to explain the variation in more than one vector. The values of the second component start to exceed one and then decline to below one (Figure 2). Therefore, the second principal component is also able to explain the two vectors. However, the third component may have a small variation, which is significant.

Thus, in the following diagram, the variable created using principal component analysis is depicted.

The supply chain activity declined from 2008. This period was during the global financial crisis. As the global economy shrunk, maritime activity should reflect the same trajectory. In 2009, the global economy began to rebound, hence, the supply chain activity (Figure 3). Then, there were two events that impacted the global economy in 2014 (Figure 3). The Ukraine–Russia conflict in 2014 led to sanctions on Russian products. This led to a fall in the supply of grain and crude oil, which increased the freight rates, thereby leading to sluggish maritime activity. And then, in December 2015, OPEC increased oil prices, which can be seen in the diagram by a fall in supply chain activity (Figure 3). The same trend in 2020 is observed. Following the COVID-19 pandemic, trade fell during the period, which led to a decline in supply chain activity (Figure 3). Furthermore, the global supply chain activity has not returned to pre-COVID-19 levels (Figure 3).

Furthermore, the following diagram depicts the dependent variable export. As mentioned above, the TAR model is estimated. Hence, the diagram depicts the dependent variable.

The period between 2009 and 2012 corresponds to regime 1, 2012 to 2020 is regime 2, and 2020 to 2023 is the third regime (Figure 4).

The other two variables used in the study are global economic activity and exchange rate depreciation. In this study, exchange rate depreciation is constructed using partial sums. This method is widely used in the literature, among others [2,30,31]. The positive component of the exchange rate represents the depreciation. Furthermore, ref. [32] conclusion shows that the impact of exchange rates is better analyzed by decomposing the variable. Below, the decomposition of the exchange rate into partial sums is shown.

Suppose a time series of ${\left\{h_t\right\}}_{t=0}^T$ is decomposed to values of the initial process.

$$h_t=h_0+h_t^{+}+h_t^{-}$$

(15)

where $h_t$ is a scalar I(1) variable, $h_0$ represents the values occurring in the beginning and $h_t^{+}$, and $h_t^{-}$ are decomposed variables. The partial sum process of the appreciation and depreciation is given by

$$h_t^{+}={\displaystyle \sum }_{i=0}^{t-1}1\left\{\Delta h_{t-1}\ge 0\right\}\Delta h_{t-1}$$

(16)

and

$$h_t^{-}={\displaystyle \sum }_{i=0}^{t-1}1\left\{\Delta h_{t-1}\ge 0\right\}\Delta h_{t-1}$$

(17)

where ${\left\{h_t^{+}\right\}}_{t=0}^T$ and ${\left\{h_t^{-}\right\}}_{t=0}^T$, respectively, denote expected appreciation and expected depreciation cumulative shocks, and both represent the initial time $t$. When the event in the parenthesis occurs, it is represented by the indicator 1; otherwise it is 0. Let us consider two integrated time series $h_{1t}$ and $h_{2t}$; both define $h_{jt}^{+}$ and $h_{jt}^{-}$ for values of $j=1, 2,$ according to $\left(16\right)$. Now assume that these series $h_{jt}^{+}$ and $h_{jt}^{-}$ are both not linearly cointergrated but within them there is a linear relationship represented by $q_t$ such that

$$b_t=x_0{h}_{1t}^{+}+x_1{h}_{1t}^{-}+x_2{h}_{2t}^{+}+x_3{h}_{2t}^{-}$$

(18)

Ref. [31] argues that $h_{1t}$ and $h_{2t}$ are asymmetrically cointegrated if there exist a vector $x^{\prime }=\left(x_0,x_1,x_2,x_3\right)$ with $x_0\ne x_1$ or $x_2\ne x_3$ $\left(\mathrm{a}\mathrm{n}\mathrm{d}{x}_0\mathrm{o}\mathrm{r}{x}_1\ne 0\mathrm{a}\mathrm{n}\mathrm{d}{x}_2\mathrm{o}\mathrm{r}{x}_3\ne 0\right)$ such that $b_t$ in $\left(D\right)$ is a stationary process. To simplify without losing generality, let us assume that only one component of each partial process is in the cointegrating relationship $\left(18\right)$, given by

$$b_t=h_{1t}^{+}-x^{+}{h}_{2t}^{+}$$

(19)

and

$$b_t=h_{1t}^{-}-{\sigma }^{-}{h}_{2t}^{-}$$

(20)

Equations $\left(19\right)$ and (20) are nonlinear. Ref. [31] argues that if a cointegrating relationship occurs in each series in $\left(18\right)$, it means there is a single direction cointegrating relationship. So, Equations $\left(19\right)$ and (20) can be rearranged

$$h_{1t}^{+}=x^{+}{h}_{2t}^{+}+b_t$$

(21)

$$h_{1t}^{-}=x^{-}{h}_{2t}^{-}+b_t$$

(22)

5. Results

The following table shows the unit root results for all the variables. Initially, the first table tests the result at the level. Furthermore, the variables are logged.

The variables measuring exports, exchange rate depreciation, and global economic activity are significant at 5%, 10%, and 5%, respectively (

Table 3. Augmented Dicky-Fuller Test.

Variables	Coefficient
exports	−3.425258 **
Supply Chain Activity	−2.247251
exchange rate depreciation	−18.27 ***
Global economic activity	−3.162 **
Supply Chain activity index	−11.04478 ***

** and ***, respectively, mean 5% and 1%. Source: Authors’ estimation.

). Meanwhile, the supply chain activity variable is insignificant (Table 3). Then, the unit root test for supply chain activity is extended by differencing the variable. The results are shown in the following table.

Following the unit root testing results at the level, the supply chain activity is tested for stationarity, and the results show the variable is significant at 1% at the first difference. Since the variables are significant at different levels, they will be converted into growth variables. This allows for the estimation to be conducted with growth variables.

Therefore, the following table shows the results of the estimated TAR model. The model has estimated three regimes where the export growth variable is the regime-switching variable. The model assumes that all the other variables are not switching. In the following table, the negative sign represents a negative growth rate of exports, and the positive sign denotes positive export growth.

The model estimated three thresholds (

Table 4. Threshold Autoregressive Model.

Variable	Coefficients
exports < −0.01939827
Supply chain activity	−0.068513 **
Global economic activity	0.000635 **
Exchange rate depreciation	−1.089402 ***
C	−0.071871 *
−0.01939827 ≤ exports < 0.03683841
Supply chain activity	0.004212
Global economic activity	0.000120
Exchange rate depreciation	0.019799
C	0.008295
0.03683841 ≤ exports
Supply chain activity	0.050213 **
Global economic activity	−0.002703 *
Exchange rate depreciation	1.792131 **
C	−0.063970 *

*, **, and ***, respectively, mean 10%, 5%, and 1%. Source: Authors’ estimation.

). In the first regime, an increase in supply chain disruptions leads to a fall in exports. These results are expected and are consistent with the literature. Ref. [8] argue that sea routes should be optimized and bigger ships should be used. In this case, when there is disruption, the impact on exports will be dampened. Whereas when global economic activity increases, exports rise (Table 4). This is expected because when a global economy expands, it is accompanied by demand for the rest of the world. Hence, the positive improvement in exports. Therefore, global economic growth improves domestic exports (Table 4). The interconnected nature of the global economic system complements improvements for all macroeconomic variables. A one-unit increase in exchange rate depreciation significantly leads to a fall in exports (Table 4). In the short run, following an exchange rate depreciation, exports worsen. This is similar to [3] findings, which used these results to support the claims of the exchange rate phenomenon.

In regime 2, all the variables are insignificant (Table 4). Regime 2 is the period after the global financial crisis of 2008. Therefore, regarding the effects of supply chain activity on exports, the insignificance impact may be due to factors such as diversification of the means of transporting goods. There may be more forms of transporting goods that dominated during this period. However, this cannot be confirmed in this study since it is beyond its scope. On the other hand, global economic activity is insignificantly related to exports. In this period, the increase in global economic activity did not improve South African exports. Lastly, exchange rate depreciation does not have an effect on export growth. This implies that this specific regime, when exchange rates depreciate, might be leading economic agents to expect them to depreciate even further. Therefore, this perception leads the economic agents to postpone their plans to demand more imports. Ref. [2] also argue that when the exchange rate depreciates, it tends to trigger further exchange rate depreciation. This bandwagon effect may be central to the inability of the exchange rate to impact exports. Furthermore, there is evidence that shows that the exchange rate of South Africa has been on a depreciation trajectory during this period [30].

In the third regime, when supply chain activity/disruption increases and exchange rate depreciation increases marginally, the exports increase. These results are similar to those obtained in regime one. The common feature of these results lies in that regime 1 is the period during the global financial crisis. The same is true for Regime 3, which is the period from the COVID-19 pandemic, which impacted the global economic system. Therefore, during turbulent economic times, the impact of these variables on exports is significant. However, in regime 3, an increase in global economic activity leads to a fall in exports. This means the response of exports in different regimes is not similar. This means that the difference in economic turmoil leads to a varying response from exports. The economic crisis captured in regime 3, restricted international trade, which affected exports. Hence, the negative effect of global economic activity on exports results from the sluggish growth of global demand and a result of slow economic activity.

6. Conclusions

In this study, the principal component analysis was used to create a supply chain activity index. The components are extracted from variables such as economic policy uncertainty, global prices of commodities, and freight rates. As shown, the first two components account for more than 80% of the variables created to capture supply chain activity/disruption. The study found that the effect of supply chain activity/disruptions on exports varies depending on the regime. In the first regime, the relationship between supply chain activity and export is negative. If there is an increase in the disruption of goods transported over the sea, the exporting countries cannot transfer the goods to other countries where they are demanded. Hence, any increase in the disturbance of sea transport leads to a decline in exports. In this case, importing economies opt for substitute goods manufactured in their home countries. Ref. [7] argued that anything leading the uncertainty about the product being delivered leads to a fall in exports. Therefore, the findings are consistent with the common theme in the literature. Furthermore, the study showed that exchange rate depreciation and export have an inverse relationship. These results imply that exchange rate depreciation results in a fall in exports. These results are not shocking because ref. [30] showed that exchange rates have a relationship that is contrary to what the economic theory predicts. Furthermore, the literature points to a negative relationship between exports and exchange rate behavior [5]. However, regime 2 corresponds to the period between two global economic crises. The relationship between export and supply chain activity and exchange rates is unclear. However, in the third regime, the relationship between the supply chain and exports is positive. This regime represents the long run, so this implies that disruptions to sea transport lead to more exports. This could be that after a long time has passed, importers use alternative routes to ensure goods are transported to the desired location. Hence, the importers demand more because they want to cushion against potential disruptions. Hence, this leads to higher exports. The relationship between exchange rates and exports is also positive in this regime. Therefore, a depreciation of the domestic currency is positively related to exports. Since the third regime represents the long run, these findings are expected. In this case, the continuous depreciation of the Rand against the Dollar leads to improvements in exports in the long run.

The limitations of this project were that the study was constrained by the availability of data prior to 2009; had it been available, the study could have revealed a broader composition of the state of affairs. Furthermore, the PCA may be considered as being based on subjective variables, so in the future, it is likely to affect the robustness of the results. Furthermore, future studies can relax the assumption that exchange rates are exogenous and consider a bidirectional relationship.