Financial Markets and the Economic Development Index in South Africa: An Econometric Approach

Abstract

Economic development is a phenomenon that involves the financial stability and standard of living of a nation’s population. To achieve economic prosperity, sound financial development, as a fundamental basis for economic development, is important. The effect of financial markets on economic development in South Africa is considered for the period 1998 to 2021. The economic development index (EDI) was used as the response variable as an indicator for economic development; financial markets were used as the explanatory variables, namely the foreign exchange (forex) markets, stock markets and money markets. The autoregressive distributed-lag econometric approach was applied. The stock market and money market were found to have a positive effect on the EDI, although only the stock market was statistically significant in terms of the probability value. The causality test showed that there exist unidirectional relationships between the stock market and the EDI; the EDI and the money market; and the forex market and the EDI. Sound financial markets and financial institutions make up a stable financial system, which makes the economy resilient to adverse shocks. Hence, unstable financial systems will have an adverse effect on the functioning of the economy by increasing the likelihood of a financial crisis.

1. Introduction

Economic development is a phenomenon that involves the financial stability and standard of living of a nation’s population. It is a process that generates the social, economic, and technological progress of a country (Coccia, 2019). When economies develop, there must be a development of the financial systems that serve the economy. This implies that for an economy to go through the process of development, the overall financial sector must be developed (Guru & Yadav, 2019). Economic development is, further, a continuous process; when some problems are resolved, new ones occur due to changes in competitive factors (Malizia et al., 2020). A country will not be able to develop if it does not have the capabilities to improve its population’s knowledge and skills, the health system, or its ability to utilise the country’s resources effectively and efficiently (Kapingura et al., 2022). To achieve economic prosperity, sound financial development, as a backbone for economic development, is important (Guan et al., 2020).

Economic development comprises numerous economic concerns, as it is about economic growth and the improvement of citizens’ welfare (Malizia et al., 2020; Suryani & Woyanti, 2021). Citizens’ welfare necessitates economic growth, which is also considered to form part of the human development dimensions. Hence, some metrics of economic development include the gross domestic product (GDP) per capita, the human development index (HDI) and the economic complexity index (ECI). The HDI is all about the nation’s social and economic dimensions, which include the standard of living, life expectancy, population literacy level, and income for various nations. The ECI captures relevant growth information, such as technology and human capital aptitudes, rather than relying on typical export diversification metrics like terms of trade shocks (Breitenbach et al., 2022). The GDP per capita is widely used to compare the economic status of nations.

The existence of a connection between financial markets and economic development has been perceived. It has been recorded in most developed economies that economic development leads to financial development (Škare et al., 2019), as well as that financial markets have a role in financing economic development plans while being a financial policy tool to leverage domestic savings and a magnet for foreign investment. South Africa has the most advanced financial markets in Africa, and they are greatly impacted by international investors. The foreign exchange (forex) market involves the purchasing and selling of national currencies. Atrill (2020) simply refers to it as a market where currencies of two different countries are exchanged. Stock market expansion is an indicator of economic development, allowing citizens access to financial resources at lower borrowing costs (Nasir et al., 2021). The money market is composed of several sub-markets, as it is not a single homogenous market; each sub-market deals with different types of short-term credit or funds (Orugun et al., 2020). Financial constraints can be reduced by well-developed financial markets, which can further increase efficiency and enhance economic development (Mamvura et al., 2020).

The novelty of this research is based on the empirical contribution, comparing how these financial markets affect South Africa’s economic development. The article is divided into several sections: the next is the literature review, followed by the methodology, the discussion of the results, and lastly the conclusion and recommendations.

2. Literature Review

2.1. Theoretical Literature

Pagano’s AK-type Endogenous growth model, and bank- and market-based financial systems are discussed.

2.1.1. Pagano’s AK-Type Endogenous Growth Model

In 1993, Pagano formalised and popularised the Endogenous growth models, further showing that economic growth is promoted by financial development when there is an efficient allocation of financial resources (Ndako, 2017; Alomari et al., 2019). Pagano developed a simple AK model which permits a significant revitalization of the analysis between growth, investment, and financial systems (stock markets, bond markets, banks, etc.) (Fanta, 2017; Bucci et al., 2020; Gaudens-Omer, 2023). The AK model is one of the new growth models which emerged as a result of the Neoclassical growth models’ shortcomings (Nwodo & Asogwa, 2017). Pagano pointed out a few pointers in which various mechanisms of finance can affect economic growth (Bucci et al., 2020). One of the internal variables influencing economic growth is the financial sector. It does so through two channels, namely, productivity and capital accumulation (Haguiga & Amani, 2019). Traditionally finance is quantified in terms of financial depth; therefore, economic growth has been found to be positively correlated with finance at lower levels of financial depth, but at higher levels growth and finance appear to be negatively correlated. This suggests that, although finance promotes growth in earlier stages of economic development, too much finance may eventually be detrimental to economic growth (Bucci et al., 2020). In the model, Pagano’s assumption is that some of the savings are lost when they are converted into investments (Gaudens-Omer, 2023). Investment has been recognised by the model as one of the key channels between financial development and economic growth (Pagano, 1993; Ndako, 2017). The equation can be formulated as

$$I=\psi E, with \psi \le 1$$

(1)

Global savings is represented by E, and the financial markets’ imperfections are reflected in the coefficient $\psi$ (Gaudens-Omer, 2023). Financial market imperfections include financial market inefficiencies, tax rates and banking intermediation margins (Nugraha, 2017; Gaudens-Omer, 2023). Financial intermediation is an expensive process that drains resources away from capital accumulation (Bucci et al., 2020). The growth model incorporates financial intermediaries through a combination of private saving, marginal capital productivity, and a fraction of saving to investment (Ndako, 2017). The economy’s growth rate is expressed as follows:

$$g={\displaystyle \frac{∆Y}{Y}}={\displaystyle \frac{AI-\delta AK}{Y}}=e\psi A-\delta$$

(2)

where $g$ is the growth rate; over time, the growth rate is constant as is customary in AK-type Endogenous growth models and depends favourably on capital productivity and the saving rate, which is reduced by the marginal rate of inefficient taxation (Nugraha, 2017; Gaudens-Omer, 2023). From the stated growth rate in Equation (2), the production function relies on capital K, in a form with constant returns ($Y=AK$) (Gaudens-Omer, 2023). $Y=AK$ implies that the aggregate output ($Y$) is the function of capital stock and factor productivity (Nwodo & Asogwa, 2017). Simply, A in Equation (2) is the productivity of capital, whereas the constant fraction of the national income $E=eY$ is represented by savings. The gross investment is stated to be the sum of the net investment $∆K$ and the replacement investment $\delta K$. Thus, $I=∆K+\delta K$ and $∆Y=A∆K$. As a result, the rate of savings and capital productivity are positively correlated with the growth of the economy, while the rates of capital depreciation and financial market defects are negatively correlated to economic growth (Gaudens-Omer, 2023). Pagano supports the view that financial development has a favourable effect on growth and development in a steady economy (Kamat, 2015).

2.1.2. Bank- and Market-Based Financial Systems

Financial systems can be classified into two groups, the market- and bank-based structure, and this varies between nations (Khoshnoud, 2016; Allen et al., 2018; Arize et al., 2018). The development of the actual economy and its institutions determines how the financial system affects economic growth. Additionally, the connection varies between times of crisis and non-crisis periods (Allen et al., 2018). Supporters of bank-based financial structures emphasise that this type of structure is effective in fostering economic growth, particularly in nations at early stages of economic development (Khoshnoud, 2016). Allen et al. (2018) assert that systems that exhibit high levels of bank financing, bank equity holding, long-term connections, close monitoring, and proactive corporate governance are known as bank-oriented systems. Financial systems that are bank-based are typically found in nations where commercial banks are widely used.

Market-based financial systems are better able to realise long-term economic growth. Countries with market-based financial systems that are well developed tend to also have bank-based financial systems that are well developed (Khoshnoud, 2016). Market-oriented financial systems facilitate the use of market-based financing by industries and support big, active securities markets. In financial systems that are market-based, banks are frequently specialised, with important groups being investment banks. In good times, market-based systems typically benefit financially dependent industries, but in bad times market-based systems tend to be detrimental (Allen et al., 2018). The argument over whether to support economic development through a market-based financial system, bank-based financial structure, or both was renewed. It was argued that, with a positive marginal productivity of capital, the expansion of financial markets causes growth and development in the short- and long-run by increasing investment efficiency. This argument was based on the principles of Endogenous growth theories (Sahoo, 2014). Growth will be slower in the absence of financial innovation; in other words, financial innovation is required to keep growth and development rates constant.

2.2. Empirical Literature

The discussion of economic development usually includes terms such as income, the labour force, wealth, and output. However, economic development also involves finances; hence, financial markets have an impact on how a country develops. Mao and An (2021) did an investigation on the connection between ECI and development levels under globalisation from 1995 to 2010, looking at middle- and high-income economies. It was discovered that increasing ECI leads to increased development in both economies. Molefhi (2021) explored the impact of macroeconomic factors on the development of capital markets, which is measured by stock markets’ and bond markets’ development in Botswana. To analyse stock market development indicators, quarterly data from 2006Q1 to 2017Q4 were used; for the bonds market development indicators, 2010Q1 to 2017Q4 data were used. The autoregressive distributed lag (ARDL) test was applied. The outcome indicated that money supply, inflation, and real output all have a favourable short-run impact on stock market development. Meanwhile, the real exchange rate has a detrimental impact. In the long run, real output was found to encourage stock market development.

Etale and Eze (2019) discussed the impact that macroeconomic variables have on the stock market in Nigeria from 1985 to 2017. The macroeconomic variables as independent variables consist of the exchange rate, inflation, interest rate and money supply. The Vector Error Correction Model (VECM) was employed. The exchange rate and money supply were found to have a considerable positive influence on the stock market, whereas inflation and interest rates had an adverse effect. Owen (2020) looked into the association between stock market development and economic growth in Nigeria from 1985 to 2018. The ARDL test was deployed, where stock markets were proxied by the value of shares traded, market-capitalisation, and turnover ratio, while economic growth was measured using the GDP. The findings revealed a short- and long-term link between the growth and development of stock markets.

Borteye and Peprah (2022) evaluated the correlation of the development of stock markets in Ghana and economic growth from 2014 to 2018. Using bivariate and regression analyses, stock market development was measured using market size, liquidity and capitalisation as the independent variables. It was discovered that there exists a high positive relationship between the development of stock markets in Ghana and economic growth. Matadeen (2019) analysed the macroeconomic and institutional elements of stock market development in Mauritius from 1989 to 2016, looking at the complementary relation between stock markets development and the banking sector. The VECM was utilised as the econometric methodology. The estimated model showed that variables were stationary when differenced once, and further that there was a cointegrating relation. Economic growth, stock market liquidity, investment and political stability all had a positive and significant effect on stock market development.

Awadzie and Garr (2020) investigated the impact of macroeconomic variables on capital market performance in Ghana. Data from 1990 to 2019 were employed, utilising econometric techniques such as the descriptive statistic, correlation matrix, and stationarity test. Stock market performance was used as a proxy for capital markets. It was found that when there is an increase in inflation rate, it positively and insignificantly affects the stock market. Meanwhile, interest rates negatively influenced the stock market, and the foreign exchange rate had a positive influence. Nigeria’s capital market and economic growth were studied by Gwaison et al. (2020) from 1981 to 2018. The ARDL bound test showed that there was a relationship among the variables. Furthermore, stock market capitalization has an insignificant positive impact on growth in the years analysed both in the short and long term. The Granger causality test was also performed, which indicated there was a unidirectional association among the variables.

Van Der Westhuizen et al. (2022) investigated the volatility transmission and interdependence between the South African stock market and the forex market from January 1979 to August 2021. The series was found to be integrated with an order of one. There was no cointegrating relationship found between the stock markets and the forex market. The empirical findings showed that there exists volatility and that the stock markets had an impact on the forex market behaviour. Furthermore, it was found that there are asymmetric spillovers and significant shocks from the forex market to the stock market. Thus, there exists a bidirectional asymmetric volatility spillover effect between the stock market and the forex market.

Matemilola et al. (2015) investigated the effect that South African monetary policy had on the bank lending rate. The study looked at the period from 1978 to December 2012. The variables used included the money market and the bank lending rate. The econometric methodologies applied were the threshold autoregressive and metric error correction models. The variables were stationarity after first difference. Cointegration between the variables is present. The momentum threshold autoregressive (M-TAR) showed that the banking lending rate and money market were cointegrated and that the adjustment process was asymmetric. The M-TAR finding further showed that the bank lending rate could adjust quicker when under the equilibrium level. Ehiedu et al. (2023) observed the kind of effect that money market instruments had on the development of the Nigerian banking sector using data from 2000 to 2018 and time series econometric techniques to attain the objectives of the study, such as the causality test and stationarity tests. The money market instruments (independent variables) included depository charges, bank acceptances and commercial paper, while the response variable was economic growth evaluated using GDP. It was revealed that the money market instruments had a long-run relationship with the Nigerian bank execution, implying that the banking sector is influenced by the money market instruments.

Kapingura (2013) did an empirical examination on the dynamic association between finance and economic growth from 1960 to 2012, looking at the country of South Africa. The econometric methodology used was the Vector Autoregression (VAR) model. A bidirectional relation between stock markets and economic growth was found, whilst there was a unidirectional causal relationship between the bonds markets and economic growth. The financial markets positively affect economic growth in South Africa. Arize et al. (2018) analysed the link between the banks and the stock market, looking at the Nigerian financial system. The ARDL test method was employed, and annual data from 1981 to 2014 were used. A long-run connection between market and bank models was found that is complementary rather than competitive, pointing to a coevolving evolution in the Nigerian financial system.

It is clear from the empirical literature that most research has not precisely focused on all the financial markets and economic development indices. Based on Pagano’s AK-type Endogenous growth model, financial markets tend to stimulate and promote economic development. There are conditions to the categorization of bank- and market-based financial systems that stem from the realisation that, in practice, the economic arrangements in individual countries are complex and vary widely, so the distinction between the two systems is not rigid. No country is a pure model, making it restrictive for making distinctions that are too sharp.

3. Research Methodology

3.1. Data

Time series and secondary annual data from 1998 to 2021 were used in this study. Economic development was captured by various indicators such as HDI, GDP per capita and ECI. Principal component analysis (PCA) was used to construct an index of economic development that is as broad as possible and captures different dimensions of the economy. The index is referred to as the economic development index (EDI). The PCA belongs to the statistical test that is part of the factor analysis (Granato et al., 2018; Khan et al., 2020). Data for HDI, ECI and GDP per capita were sourced from our World in data, the Observatory and Atlas Economic complexity index and the Word Bank databases. HDI is one of the best statistical tools that enables a country to keep track of its level of development. The ECI can forecast important macroeconomic effects, such as the degree of wealth in a society, economic advancement, and income inequality (Li et al., 2021). GDP per capita is a key metric for assessing a nation’s economic success, which is a major factor in determining the economic well-being of its citizens.

Data for the financial market (stock market, forex market and money market) variables were obtained from the World Bank. The stock market return data was used, which is the yearly growth rate of the stock market index, as it can act as a key barometer for overall market performance, investor sentiment and economic health. It is evident from the theories adopted in this study that, indeed, financial markets are expected to positively influence and lead to economic development.

3.2. Model Specification

In the estimated model, economic development is a function of financial markets. The estimated model, in linear form, is as follows:

$${EDI}_t={\beta }_0+{\beta }_1{StockMINDX}_t+{\beta }_2{MoneyM}_t+{\beta }_3{ForexM}_t+{\epsilon }_t$$

(3)

The linear log form of the models stated above is as follows:

$${EDI}_t={\beta }_0+{\beta }_1{StockMINDX}_t+{\beta }_2{LMoneyM}_t+{\beta }_3{LForexM}_t+{\epsilon }_t$$

(4)

where $StockMINDX$ represents the stock market index, the logged money market is shown by $LMoneyM$, and $LForexM$ is the logged forex market. ${\beta }_0$ is the constant and ${\beta }_1$, ${\beta }_2$ and ${\beta }_3$ represent the coefficient estimates. $\epsilon$ represents the error term and $t$ is the time trend.

3.3. Estimation Techniques

The different econometric techniques conducted, namely the stationarity tests, ARDL bounds test, the Granger causality test, and lastly the diagnostic tests, are discussed. The PCA was also conducted, as it was used to capture all aspects of economic development. The PCA is increasingly employed in welfare measures and is used to obtain coefficients that provide correct weights based on the statistical significance of each included variable in the index. By identifying these new variables, the principal component (PC) solves the eigenvalue/eigenvector.

3.3.1. Descriptive Statistic

The descriptive statistic shows the mean, median, maximum, minimum, standard deviation, skewness, Kurtosis, Jarque–Bera and probability (p-value) (Mishra et al., 2019; Koondhar et al., 2021). The mean refers to the data’s mathematical average value. The median is described as the middle observation when data are ordered in either a decreasing or increasing order of magnitude. The observation that has maximum frequency is referred to as the mode, as it is the value in a set of observations that occurs most frequently. The standard deviation (SD) measures how spread out the values are from its mean value (Gashiten, 2023).

3.3.2. Correlation Matrix

The correlation matrix establishes the linear relationship between the variables used. When data are of high dimension and might not be normally distributed, the correlation test for two or more vectors is proposed to be conducted (Ahmad & Ahmed, 2021).

3.3.3. Stationarity

The traditional stationarity tests are conducted, namely the Augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) tests proposed by Dickey and Fuller (1979) and Phillips and Perron (1988). If the $x$ and $y$ variables are nonstationary, the spurious regression can be generated when modelling the $x$ and $y$ variables’ relationship as a simple OLS relationship, as in the below equation:

$$Y_t=\alpha +{\beta X}_t+{\epsilon }_t$$

(5)

The stationarity test is usually performed to make sure that the tested variables are not integrated at a higher order than one. The series is stationary if both the mean and variance are constant. Otherwise, the series will be nonstationary. When the series is nonstationary at levels, the differencing process occurs.

$$x \mathrm{Level} x_t$$

(6)

$$x \mathrm{First} \mathrm{difference} x_t-x_{t-1}$$

(7)

$$x \mathrm{Second} \mathrm{difference} x_t-x_{t-2}$$

(8)

When a series is stationary at levels, without differencing, it is stated to be of $I\left(0\right)$. If the series is stationary at first difference or second difference, it is stated to be of $I\left(1\right)$ or $I\left(2\right)$.

3.3.4. ARDL Bounds Test

An ARDL bounds test is appropriate for a model that is integrated at different orders, such as order zero or one. The ARDL model is more advanced regardless of the sample size used (Owen, 2020). Modelling the ARDL with suitable lags is standard for both the serial correlation and endogeneity problem (Owen, 2020; Molefhi, 2021). The ARDL bound test can be used to determine the short- and long-run relation between the financial markets and EDI, presenting unbiased estimates, as the short- and long-run estimates can be determined at the same time, where problems associated with autocorrelation and omitted variables are eliminated (Molefhi, 2021). The simplified ARDL model can be expressed as follows:

$$∆y_t={\beta }_1+{\delta }_1{y}_{t-1}+{\delta }_2{x}_{t-1}+{\delta }_3{z}_{t-1}+{\eta }_1{\sum }_{i=1}^n∆y_{t-1}+{\eta }_2{\sum }_{i=1}^n∆x_{t-1}+{\eta }_3{\sum }_{i=1}^n∆z_{t-1}+{\mu }_t$$

(9)

where its sum corresponds to that of the error correction model (ECM). ${\eta }_1$ to ${\eta }_3$ denote the short-run coefficients and ${\mu }_t$ is the stochastic term. The generalised model for examining the effects of financial markets on economic development is specified as

$$\begin{array}{l}∆({EDI)}_t={{\alpha }_0+\beta }_1∆(StockMINDX{)}_{t-1}+{\beta }_2{∆l\left(MoneyM\right)}_{t-1}+{\beta }_3{∆l\left(ForexM\right)}_{t-1} +\\ {\varnothing }_4{∆log\left(StockMINDX\right)}_{t-1}+{\varnothing }_5{∆log\left(MoneyM\right)}_{t-1}+{\varnothing }_6{∆log\left(ForexM\right)}_{t-1}+{\mu }_t\end{array}$$

(10)

In Equation (10), $t-1$ represents the lagged period, $∆$ is the differencing of the variables, and lastly ${\varnothing }_4$ to ${\varnothing }_6$ are the long-run coefficients (Owen, 2020).

Since the ARDL test also checks if there is cointegration in the series, the F-statistic is used. The decision rule that applies to the acceptance and rejection of the null hypothesis is that, firstly, if the F-statistic is greater than the lower and upper bound critical values, it means there exists cointegration. Therefore, the null hypothesis of no cointegration between economic development and financial markets will be rejected and the alternative hypothesis accepted. Secondly, if the F-statistic is less than the lower and upper bound critical values, it means there is no cointegration. In this case, the null hypothesis of no cointegration between financial markets and economic development will be accepted. Lastly, if the F-statistic lies between the lower and upper bound critical values, it means that the cointegration decision is inconclusive. This decision’s rules are in line with Pesaran et al. (2001).

The speed of adjustment or the unrestricted ECM can also be determined under the ARDL bound test approach to check the long-run equilibrium, where the response and independent variables are used to estimate the best fit model (Owen, 2020).

3.3.5. Granger Causality Test

Granger causality, according to Gujarati (2004), takes place when there is a correlation only amongst the current value of one variable and the previous values of other variables. In a bivariate ($x,y$) context, a general specification of the Granger causality can be expressed as

$$Y_t={\alpha }_0+{\alpha }_1{Y}_{t-1}+\dots +{\alpha }_i{Y}_{t-i}+{\beta }_1{X}_{t-1}+\dots {\beta }_i{X}_{t-i}+\mu$$

(11)

$$X_t={\alpha }_0+{\alpha }_1{X}_{t-1}+\dots +{\alpha }_i{X}_{t-i}+{\beta }_1{Y}_{t-1}+\dots {\beta }_i{Y}_{t-i}+\mu$$

(12)

In the equations above, the subscripts represent the time periods, and white noise error is denoted by $\mu$. The constant growth rate of $y$ in Equation (11) and $x$ in Equation (12) is represented by the constant parameter “0”. The test firstly estimates the null hypothesis that the explanatory variable ($X$) does not Granger-cause the response variable ($Y$) and secondly the null hypothesis that the response variable ($Y$) does not Granger-cause the explanatory variable ($X$). If the former null hypothesis is not rejected and the latter rejected, it can be concluded that changes in the explanatory variable are Granger-caused by changes in the response variable. This further means that there is a unidirectional causal link among the two variables. A bidirectional causal relationship occurs when both null hypotheses are rejected.

3.3.6. Diagnostic Tests

Assessing data normality is a prerequisite for many statistical tests; the Jarque–Bera normality test is used. In many modelling frameworks, normality is also a useful assumption, including general linear models that are well known to assume residuals that are normally distributed, and structural equation modelling, where the common starting point is the normal theory-based maximum likelihood estimation. The Breusch–Pagan test can be performed to test the performance of the individual effects (Gouvêa & Lima, 2013; Ceesay & Moussa, 2022). The Breusch–Pagan test is a Lagrange multiplier (LM) test, with a null hypothesis in which the individual effects variance is zero.

The heteroskedasticity tests are used to check if the model or variables are free from the issue of heteroskedasticity (Efanga et al., 2020). Nkwede (2020) stipulates that the heteroskedasticity test detects whether an association exists in the estimated model between the independent variables and residual value and ascertains that there is no violation of the constant variance assumption, which might lead to the dilemma of heteroskedasticity.

4. Results and Discussion

The findings from the PCA, descriptive statistic, correlation matrix, stationarity test, ARDL test, Granger causality, and diagnostic tests are presented in this section.

4.1. EDI Construction Through the PCA

The PCA results are thus incorporated and presented first. According to Ali et al. (2022), the principal components (PCs) with the highest eigenvalues that are greater than 1 must be considered.

In

Table 1. Construction of EDI by the PCA.

Eigenvalues: (Sum = 3, Average = 1)
Number	Value	Difference	Proportion	Cumulative Value	Cumulative Proportion
1	2.392283	1.912492	0.7974	2.392283	0.7974
2	0.479791	0.351864	0.1599	2.872073	0.9574
3	0.127927	---	0.0426	3.000000	1.0000
Eigenvectors (Loadings):
Development Indicators	PC 1	PC 2	PC 3
HDI	0.620300	−0.048079	0.782890
GDP Per Capita	0.547486	0.741292	−0.388259
ECI	−0.561683	0.669458	0.486146

Author’s own computation.

, the first summarised results are the eigenvalues, whose information shows that the first principal has a higher value of 80%, while the second is at 16%. The cumulative proportion of information that is explained by the first two principal directions is around 96%.

4.2. Descriptive Statistic Results

To rule out the outlier’s possibility in the data, the minimum and maximum statistics are used (Awadzie & Garr, 2020). Furthermore, it is required that the model is normally distributed as with classical linear regression.

It is evident from the results of the descriptive statistics in

Table 2. Descriptive statistic results.

	EDI	StockMINDX	LMoneyM	LForexM
Mean	7.44 × 10−15	11.33	1.79	0.96
Median	−0.08	11.26	1.82	0.92
Maximum	1.01	43.66	1.87	1.22
Minimum	−1.41	−13.60	1.68	0.74
Std. Dev.	0.71	14.95	0.06	0.14
Skewness	−0.28	0.19	−0.74	0.34
Kurtosis	2.00	2.43	2.11	1.76
Jarque–Bera	1.30	0.48	3.00	2.02
p-Value	0.52	0.79	0.22	0.36

Author’s own computation.

that the EDI has an average return of around 0% and a standard deviation at 0.71, with values lying between −1.41% and 1.01%. The stock market has a better performance, with an average return of 11% and standard deviation of 14.95, with values lying between −13.6% and 43.66%. The money markets indicate a mean value of 1.79% and standard deviation of 0.06, with values lying between 1.68% and 1.87%. The forex market has a mean value of 0.96% and standard deviation of 0.14, with the values lying between 0.74% and 1.22%. The stock market’s performance is followed by the money markets and forex market. The EDI and money markets are found to be negatively skewed, unlike the stock market and forex market, which are positively skewed. All the variables are not highly peaked, as the Kurtosis values are less than 3.

4.3. Correlation Matrix Results

To estimate the direction and strength of the linear link between the factors, and also to ascertain if there exists multicollinearity among the variables, a correlation matrix was conducted. The outcomes are shown in

Table 3. Correlation matrix results.

	EDI	StockMINDX	LMoneyM	LForexM
EDI	1	0.2563	0.4243	−0.2865
StockMINDX	0.2563	1	−0.0375	−0.3001
LMoneyM	0.4243	−0.0375	1	0.5082
LForexM	−0.2865	−0.3001	0.5082	1

Author’s own computation.

below.

The correlation results indicate that EDI is positively correlated to the stock markets and money markets, but negatively correlated to the forex market. The EDI and the stock market have a positive correlation of 0.26, which shows that a percentage increase in the stock markets will lead to an increase of 26% in the EDI. This finding is supported by Adoms et al. (2020) and Salameh and Ahmad (2022), who proffer that stock markets are correlated with economic development. The EDI and the money market also have a positive correlation of 0.42, implying that a percentage increase in the money markets will lead to a 42% increase in the EDI. The EDI and forex markets have a negative correlation of 0.29, which reveals that a percentage increase in the forex markets will result in a 29% decline in the EDI. The correlation is low and shows that the sensitive development variables used in the model will partially account for the differences in the forex markets at the various levels of development (Roncaglia de Carvalho et al., 2018). The money market is the indicator with a greater and positive impact on the EDI, followed by the stock markets.

4.4. Stationarity Results

The formal unit root results are provided in the following

Table 4. Formal stationarity results.

		Stationarity at I(0)		Stationarity at I(1)
Variables	Model Specification	ADF Test Statistic	PP Test Statistic	ADF Test Statistic	PP Test Statistic
EDI	Individual intercept	−1.320623	−1.208578	−6.328434 ***	−6.111287 ***
	Individual intercept and trend	−1.187608	−0.958166	−7.915934 ***	−8.413453 ***
	None	−1.369490	−1.262276	−6.518751 ***	−6.277393 ***
StockMINDX	Individual intercept	−3.571169 **	−3.573124 **
	Individual intercept and trend	−3.695121 **	−3.591292 *
	None	−2.361955 **	−2.407208 **
LMoneyM	Individual intercept	−1.504969	−1.510359	−3.733323 **	−3.724281 **
	Individual intercept and trend	−1.261893	−1.352962	−3.775161 **	−3.775161 **
	None	1.491892	1.331397	−3.538168 ***	−3.526460 ***
LForexM	Individual intercept	−1.396727	−1.179511	−3.329227 **	−3.054424 **
	Individual intercept and trend	−2.778645	−1.867982	−3.721186 **	−3.294034 *
	None	1.038587	1.217897	−3.249026 ***	−3.034901 ***

Author’s own computation. The asterisk indicates * rejection of the null hypothesis at 10%, ** at 5%, and *** at 1%.

.

The formal unit root results in Table 4 indicate that there is stationarity in the model. The ADF and PP test statistics indicate that, at levels, for the EDI, there is a unit root under all testing procedures. To induce stationarity in these three procedures, the EDI was differenced. This resulted in the rejection of the null hypothesis at 1%, and the EDI is found to be $I\left(1\right)$. The ADF and PP t-statistics outcome further showed that, at levels, there is no unit root for the stock market, and the stock market is $I\left(0\right)$.

The money market and forex market were found not to be stationary at levels. The money and forex markets had to be differenced once to be stationary at all testing procedures. After first differencing, the null hypothesis was rejected at 1%, 5% and 10% significance levels, and the alternative hypothesis that there is stationarity was accepted. Overall, some of the variables were not differenced, and others were differenced once to induce stationarity. From the findings, it can be stated that the series is a mixture of $I\left(0\right)$ and $I\left(1\right)$. Adoms et al. (2020) stipulate that for the ARDL test to be conducted the variables used in the model must be $I\left(0\right)$ or $I\left(1\right)$. To distinguish a relationship between the variables, the ARDL model is used.

4.5. ARDL Bounds Test Results

The ARDL provides the long-run and bound test results. These results reveal whether there is a long- or short-run cointegrating relationship among the variables used in the model. Furthermore, they reveal if the series will converge back to equilibrium. The variables are $I\left(0\right)$ and $I\left(1\right)$ in the stationarity test, and such results allow the ARDL bound test to be conducted. The F-bound ARDL cointegration test results are presented first, in

Table 5. ARDL bound results.

Test Statistic	Value	Significance	Critical Bounds Values
			I(0)	I(1)
F-statistic	5.132778	10%	2.37	3.2
k	3	5%	2.79	3.67
		2.5%	3.15	4.08
		1%	3.65	4.66

Author’s own computations.

. The null hypothesis of the test is that there is no cointegrating relationship among the variables, namely the EDI and financial markets, while the alternative hypothesis is that there exists a cointegrating relationship between the EDI and financial markets.

The F-bound ARDL results in Table 5 show that the F-statistic value of 5.132778 is greater than all the critical bound values at both the lower [$I\left(0\right)$] and upper bounds [$I\left(1\right)$] at all significance levels. This outcome reveals that the variables (EDI, stock market, money market and forex market) are cointegrated. This is similar to the results of Gwaison et al. (2020), who found a long-run cointegrating relationship among the variables. The ARDL long-run outcomes are provided below in

Table 6. ARDL long-run results.

	Coefficient	p-Value
StockMINDX	0.076699	0.0721
LMoneyM	5.692941	0.1814
LForexM	−2.230848	0.2084
C	−8.811124	0.1990

Author’s own computations.

.

The following long-run equation is derived from the estimated parameters:

$${EDI}_t=0.076699{StockMINDX}_t+5.692941{LMoneyM}_t -2.230848{LForexM}_t+{\epsilon }_t$$

(13)

The long-run Equation (13), derived from the ARDL results, indicates that the stock market and money market have a positive influence on the EDI in South Africa. The finding meets the prior expectations of the study, except for the forex market. However, this relationship is statistically significant for the stock market and not the money market. The stock market has a probability value (p-value) of 0.0721, which is less than the 0.1 significance level. The stock markets having a positive and significant influence on the EDI implies that a 10% increase in the stock market will lead to a 0.76% increase in the EDI. The studies of Gwaison et al. (2020) and Borteye and Peprah (2022) also revealed a long-run relationship between the stock market and economic development. These outcomes show that, as stock markets develop, economic development will improve both in the short and long run in South Africa. Stock markets have been turned to as a form of avenue to provide firms, companies, and governments with capital to be able to fund projects instead of relying on other financial institutions (Molefhi, 2021). The outcomes further confirm that a 10% increase in the money markets will lead to a 57% increase in the EDI. The money market has an insignificant p-value of 0.1814, which is greater than the 0.1 significance level. The money markets bear the greatest percentage in influencing the EDI, followed by the forex market. The money market is a real means of attaining economic development, although it was found to have an insignificant p-value in the study. The positive effect of the stock and money markets is consistent with Kapingura (2013), who stipulated that financial markets in South Africa tend to positively influence economic development.

The forex market bears a negative sign; a 10% increase in the forex will lead to a 22% decline in the EDI. However, the forex market, similar to the money market, does not possess a significant p-value. Maintaining competitive and stable exchange rates can spur economic development; yet, because of the volatility of the world’s financial markets, this calls for adaptable and persistent interventions (Guzman et al., 2018). It is evident from the long-run ARDL results that some financial markets do have a positive effect on economic development in South Africa.

From the short-run results in

Table 7. ARDL short-run (ECM regression) results.

	Coefficient	p-Value
D(StockMINDX)	0.011803	0.0085
D(LMoneyM)	−8.186425	0.0267
LForexM	−0.872099	0.2909
Speed of adjustment (ECT): −0.390927		0.0000

Author’s own computation.

, the stock market positively and statistically significantly influences the EDI, unlike the money market, which is negative but statistically significant. In terms of the long-run equilibrium, the path of the economy can be characterised by dynamic econometric models such as the ARDL model (Colmenares et al., 2021). The speed of adjustment is found to be −0.390927. The speed of adjustment indicates whether the model will converge to equilibrium or not. The outcome indicates that about 39% of errors in the short run are corrected back to equilibrium in the long run. The economic development in South Africa will converge back to its long-run equilibrium path in response to changes in its regressors, namely the stock markets, money markets and forex markets. This shows that financial markets have an effect on economic development, both in the short and long run, in South Africa.

4.6. Granger Causality Results

The causality test was utilised to analyse the presence of the two-way association among the variables. The outcomes from the test are presented in

Table 8. Granger causality.

Null Hypothesis	p-Value
StockMINDX does not Granger-cause EDI	0.0252
EDI does not Granger-cause StockMINDX	0.7921
LMoneyM does not Granger-cause EDI	0.1671
EDI does not Granger-cause LMoneyM	0.0008
LForexM does not Granger-cause EDI	0.0095
EDI does not Granger-cause LForexM	0.7886

Author’s own computations.

.

From the Granger causality results, the stock market does Granger-cause the EDI and the EDI does not Granger-cause the stock market. This indicates that any changes in the stock market do influence economic development. There is a unidirectional causal relationship running from the stock market to the EDI. In the study of Kapingura (2013), the results indicated a bidirectional relationship. The p-values are 0.0252, which is less than the 0.05 significance level, and 0.7921, which is greater than the 0.05 significance level. The finding implies that there is a rejection of the null hypothesis that the stock market does not Granger-cause the EDI. The money market does not Granger-cause the EDI, while the EDI does Granger-cause the money market. The null hypothesis that the EDI does not Granger-cause the money market is rejected in this case at a 5% significance level, as the p-value of 0.0008 is less than 0.05. This further indicates a unidirectional relationship running from the EDI to the money market. The forex market does Granger-cause the EDI, while the EDI does not Granger-cause the forex market. The null hypothesis that the forex market does not Granger-cause the EDI is rejected at a 5% significance level. The p-value of 0.0095 is less than 0.05. There is a unidirectional relationship between the two variables.

From the Granger causality results, it is evident that there is a unidirectional causal relationship between the stock markets and the EDI, the EDI and the money market, and the forex market and the EDI. A bidirectional relationship between the variables was not found. These outcomes show that changes in some financial markets do influence economic development in South Africa.

4.7. Diagnostic Results

To evaluate the validity of the model, diagnostic tests were conducted.

Table 9. Diagnostic results.

Techniques	H0	Statistic	p-Value
Jarque–Bera normality test	Normally distributed	0.896852 (Jarque–Bera)	0.638632
Breusch–Pagan–Godfrey: serial correlation LM	No serial correlation	0.093424 (Obs*R-sqaured)	0.9719
Heteroskedasticity test: Breusch–Pagan–Godfrey	No heteroskedasticity	1.236460 (Obs*R-sqaured)	0.7810
Heteroskedasticity test: ARCH	No heteroskedasticity	0.471244 (Obs*R-sqaured)	0.5147

Author’s own computations. Note: Obs* represents number of observations multiplied by the R-squared.

provides the diagnostic test results. The Jarque–Bera statistic is about 0.896852, and the probability of obtaining such under-normality assumptions is about 64%. The serial correlation test has an observed R-squared of 0.093424 and a p-value of 0.9719, which exceeds 0.05. The Breusch–Pagan–Godfrey heteroskedasticity test has an observed R-squared of 1.236460 and a p-value of 0.7810, which exceeds the level of significance of 0.05. Like the Breusch–Pagan–Godfrey heteroskedasticity test, the ARCH heteroskedasticity test shows an observed R-squared of 0.471244 and a p-value of 0.5147, which also exceeds 0.05. It is evident from the diagnostic results that $H_0$ must be accepted, as the model is normally distributed and free from serial correlation and heteroskedasctity problems.

5. Conclusions and Recommendation

This study has given an account to the effect of financial markets on the economic development index in South Africa, using secondary annual data from 1998 to 2021. The study applied various economic theories to provide a broader discussion of the variables and to achieve the objective of the study. The objective of the study was to determine the effect of financial markets on economic development in South Africa. The ARDL bound test was employed as the econometric methodology. The results in the EDI model showed that the variables are integrated with an order of zero and one, implying that the model is stationary. The cointegration test showed that the variables move together in the long-run and are correlated. The ARDL long-run outcome showed that the EDI is positively and significantly influenced by the stock markets. It was evident that financial markets had a positive effect on economic development in South Africa. A unidirectional relationship among the stock market and the EDI, the EDI and the money market, and the forex market and the EDI was found. The residuals were normally distributed, with no heteroskedasticity and no serial correlation, and the series was of good fit. The model is effectively identified and can be used to forecast future movements.

There should be sustainable qualitative and quantitative improvements for economic development to occur. A country must have effective economic strategies to experience improvements in the development of its economy. There should be stability of exchange rates, including the money supply, while at the same time keeping interest rates and inflation trends as stable as possible so the performance of financial markets can be enhanced, thus bringing about economic development. Lower rates of inflation, money supply injections and developed financial markets in a country where there is a higher lifespan and education level make it possible for economic development to be achieved. Developing and developed financial markets, in emerging market economies, are expected to play a greater role in driving economic development. Financial markets can further boost economic development in South Africa by promoting the movement and flow of funds to deficit units from the saving surplus. Sound financial markets and financial institutions make up a stable financial system, which makes the economy resilient to adverse shocks. Hence, a financial system that is not stable will have a negative effect on the performance of the economy by increasing the likelihood of a financial crisis. Thus, the development of well-functioning financial markets is important. Financial market participants, policymakers, monetary authorities, and central banks must ensure that there is accurate and reliable information when it comes to the management of crisis conditions within the economy.

6. Contribution and Limitations of the Study

Issues of development have received attention in academic and political debates internationally, in Africa and in South Africa. Development studies, practices and policies have been thriving over the years, as this field of study has become one of the most important research fields in economics (Coccia, 2019). Most studies have focused on the relation between financial development and economic growth; the inequality of economic development; financial development and economic development; financial development, economic openness, and economic progress; and financial markets and the COVID-19 pandemic (Arif & Khan, 2019; Škare et al., 2019; Farouq et al., 2020; Sansa, 2020; Şenol & Zeren, 2020; Hussain et al., 2021; Antara, 2022; Kapalu & Kodongo, 2022). Looking at the existing literature, a gap exists in the available research. This study attempted to bring a different perspective on numerous issues at hand by focusing on how, econometrically, economic development is affected by financial markets in South Africa. This study contributes to economic policies and will inform policy-makers and governments on how to improve economic development policies through financial markets. The growing importance of financial markets has strengthened the general conviction that finance is a significant factor of economic development.

There were potential limitations in terms of the availability of data; hence, the study mainly focused on the years 1998 to 2021.