Is stock market development a spur to poverty reduction? Empirical evidence from selected sub-Saharan African countries

ABSTRACT In this study, the relationship between stock market development and poverty reduction is examined in sub-Saharan African countries during the period 1993–2019. Three proxies of stock market development were used, namely stock market capitalization, stock market traded value and stock market turnover ratio, thereby leading to three separate model specifications. A battery of panel data estimation techniques was used in the study to examine the existence of cross-sectional dependence. Furthermore, the second-generation panel unit root test and panel cointegration test were used alongside the first-generation tests to examine this linkage. Contrary to some of the previous studies, the results of the panel least squares, fully modified ordinary least squares and dynamic ordinary least squares show that stock market development has no significant impact on poverty reduction in sub-Saharan countries during the period under study. These results apply irrespective of the proxy used to measure the level of stock market development. The results were also corroborated by the heterogenous non-Granger causality test, in which a prima facie causal flow was found from poverty reduction to stock market development when stock market turnover ratio was used as a proxy. Policy implications are discussed.


Introduction
The role of financial development in poverty reduction has attracted attention in recent years on both the theoretical and empirical fonts.Previous studies have shown that financial sector development can have an impact on poverty through two channels, namely (i) direct channel when the poor benefits from financial development directly through easy access to financial services; and (ii) indirect channel, where financial development leads to poverty reduction through economic growth (see Zhuang et al. 2009;Claessens and Feijen 2006).The former is transmitted through (i) the expansion of job opportunities in the sector such as small and medium-sized enterprises; (ii) increased investment in human capital; and (iii) Consumption smoothing (see Claessens and Feijen 2006).The latter, on the other hand, is transmitted through the so-called trickle-down effect.Given the presumed positive role of financial development on economic growth, it has been argued that financial development could trickle down to the poor through its supply-leading response (see Odhiambo 2009; 2011 for more details).
Although some researchers have argued that financial development is pro-poor as it leads to poverty reduction, some still argue that the beneficial effects of financial development may not trickle down to the poorest of the poor.In some studies, it has even been argued that financial development trickles up to the middle class and the super-rich (see Odhiambo 2011).The dominant argument, however, has been that a well-developed financial sector could result in poverty reduction by granting the poor access to formal finance (see also Jalilian and Kirkpatrick 2001;Odhiambo 2010Odhiambo , 2013)).
In some studies, it has also been shown that financial development enables the poor to draw down their accumulated savings and even borrow to start their own micro-enterprises (see Odhiambo 2013).Access to loanable funds for investment by the poor leads to higher income, more employment opportunities and poverty reduction (see DFID 2004;Odhiambo 2009Odhiambo , 2013)).In other studies, it has been shown that lowering the high fixed cost of lending to small borrowers could enable the poor to access formal finance from the credit market (see Stiglitz 1998;Jalilian and Kirkpatrick 2001).By relaxing the liquidity constraints inherent in some financial systems in developing countries, financial development gives the poor access to the scarce loanable funds necessary for investment, employment and poverty reduction.In previous studies, it has also been shown that improved financial services enables entrepreneurs and households to reduce the cost of managing their risks, thereby expanding their economic opportunities (see Atkinson and Stiglitz 1980;Townsend 1982;Bardhan, Bowles, and Gintis 2000). 1  Unfortunately, in the presence of imperfect credit market and financial market failures, which are prevalent in underdeveloped financial markets, poorer households may not access loanable funds from the financial markets.This could have a negative effect on investment in human and physical capital, economic growth and poverty reduction.Moreover, in some studies, it has been shown that weak financial systems may transfer funds to the rich and elites with strong political connections, thereby omitting the bulk of the poorer population (see Rajan and Zingales 2003;Claessens and Perotti 2007). 2 The impact of financial development on poverty reduction, therefore, largely depends on the nature and strength of the financial market.
Although the role of stock market development has been investigated extensively in many countries, the majority of previous studies have focused largely on the link between stock market development and economic growth (see van Nieuwerburgh, Buelens, and Cuyvers 2006;Caporale, Howells, Soliman 2004).Indeed, the bulk of the literature on the link between financial development and poverty reduction in many developing countries has been dominated by the role of bank-based financial development, rather than stock market development.According to Kpodar and Singh (2011), when institutions are weak, bank-based financial systems are better at reducing poverty than market-based financial systems.However, as institutions grow and become more developed, market-based financial systems become more effective in the reduction of poverty.
While the financial sector in sub-Sharan Africa (SSA) has deepened significantly in recent years, it is still lagging behind many other regions globally.In particular, the region has led the world in innovative financial services based on mobile telephony, such as M-Pesa, M-Shwari and M-Kopa in Kenya, which has significantly reduced the transaction costs in areas where there is inadequate financial infrastructure (see IMF 2016).However, unlike other regions where both the bankbased and market-based financial systems have thrived, the financial systems in many SSA countries are largely dominated by bank-based financial systems.Even within the banking system, foreign-owned subsidiary banks tend to account for a major share of the financial assets in countries such as Guinea, Guinea-Bissau, Madagascar, São Tomé and Príncipe, while in countries such as Ethiopia, Rwanda, the Seychelles and Sierra Leone, the assets of state-owned banks are sizable (see IMF 2016).
Although the region's median ratio of private sector credit to GDP increased significantly in 2014 to 21%, it remained relatively low when compared to other regions, such as the Middle East and North Africa (MENA), East Asia, and Latin America and the Caribbean (IMF 2016).According to the IMF (2016), this low median ratio of private sector credit to GDP is largely driven by the region's high number of low-income countries.According to the World Bank's (2021), 23 of the 27 low-income countries are currently in the SSA region.In the main, the financial system in SSA is dominated by commercial banks with very few investment banks (see Soumaré et al. 2021).Despite the fact that the capital markets in many SSA countries are still underdeveloped, the Johannesburg Stock Exchange of South Africa remains not only the largest stock exchange in Africa, but also one of the largest in the world by market capitalization.As of 2018, the Johannesburg stock exchange market capitalization as a percentage of GDP was estimated to be 352%.Unfortunately, the majority of previous studies on the relationship between financial development and poverty in SSA countries have mainly focussed on bank-based financial development, rather than stock market development (see Odhiambo and Zerihun 2019;Acheampong et al. 2021).Others focussed on political regimes and stock market performance in Africa (Asongu and Nwachukwu 2018), stock market convergence (Asongu, 2013a).In studies where the link between stock market and poverty reduction has been examined in developing countries, the main focus has been on Asia and Latin America, leaving SSA countries with little or no coverage (Uddin et al. 2014;Lazar, Priya, and Jeyapaul 2006).
It is against this background that the current study aims to examine the relationship between stock market development and poverty reduction in SSA countries where sufficient data exist using novel econometric techniques.The current study differs fundamentally from some of the previous studies done on other countries in two ways.First, three proxies of stock market development were used, namely stock market capitalization, stock market traded value ratio and stock market turnover ratio, thereby leading to three different specifications.Secondly, a battery of modern panel data techniques has been employed in the study in a stepwise fashion to examine this linkage.These include, among others, (1) cross-sectional dependence tests based on Breusch and Pagan (1980) LM; Pesaran (2004) scaled LM; Baltagi, Feng, and Kao (2012) bias-corrected scaled LM; and Pesaran (2004) Dumitrescu and Hurlin (2012).To our knowledge, this may be the first study to fully examine the dynamic linkage between stock market development and poverty reduction in SSA on this scale.
The rest of the paper is organized as follows: In Section 2, the literature on the relationship between financial development and poverty reduction is reviewed in developed and developing countries.In Section 3, the methodology employed, and the estimation techniques used are presented.In Section 4, the empirical analysis and the discussion of the results are presented, while the study is concluded in Section 5.

Empirical literature review
Theoretically, financial development can lead to a reduction in poverty through two main channels (see Zhuang et al. 2009).The first channel, which is referred to as trickle-down channels is via economic growth.A well-developed financial sector that is wide, deep and efficient, is expected to facilitate the process of capital accumulation, which leads to an increase in economic growth (see also Abosedra, Shahbaz, and Nawaz 2016).Through its influence on economic growth, financial development is, therefore, expected to trickle down to the poor and poorest through job creation and other economic opportunities (Zhuang et al. 2009).In some studies, it has also been shown that higher economic growth, which results from financial development, is likely to benefit the poor through its impact on narrowing the wage differentials between skilled and unskilled labor (Galor and Tsiddon 1996). 3In addition, an increase in economic growth due to the beneficial effects of financial sector development is likely to lead to an increase in tax revenue and government spending in social services, such as health, education and social protection, which largely tend to benefit the poor (see Abosedra, Shahbaz, and Nawaz 2016).The second channel through which a well-developed financial sector could benefit the poor is through its positive impact on financial access to the poor.A well-developed, efficient financial sector is likely to give the poor easy access to financial services, such as credit and insurance risk services, which eventually boosts their productivity and economic welfare (Abosedra, Shahbaz, and Nawaz 2016;World Bank 2001).A more developed, competitive financial sector is also able to provide better financial services, products and rates, which enhances the quality of the poor's living standards, thereby enabling them to get out of the perpetual cycle of poverty (Beck, Demirgüç-Kunt, and Levine 2007).
On the empirical front, the relationship between financial development and poverty reduction can be clustered into two strands.The first strand argues that financial development is pro-poor and has an unambiguous positive effect on poverty reduction.Studies, in which the findings are consistent with this view include those conducted by Beck, Demirguc-Kunt, and Levine (2004), Honohan (2004), Lazar, Priya, and Jeyapaul (2006) for the case of market capitalization ratio and concentration, Beck, Demirgüç-Kunt, and Levine (2007), Honohan and Beck (2007), Jeanneney and Kpodar (2011), Beck and Maimbo (2013), Uddin et al. (2014), Abosedra, Shahbaz, and Nawaz (2016), Bayar (2017), Cepparulo, Cuestas, and Intartaglia (2017), Rewilak (2017), Sehrawat and Giri (2018), and Appiah, Frowne, and Tetteh (2020), among others.Beck, Demirguc-Kunt, and Levine (2004), for example, found that financial development reduces income inequality by disproportionately boosting the incomes of the poor.Honohan (2004), while examining the link between financial development, growth and poverty, found that a 10-percentage point increase in the ratio of private credit to GDP should reduce poverty rations by 2.5 to 3 percentage points.Lazar, Priya, and Jeyapaul (2006), while examining the relationship between capital market and poverty reduction in India, found that in some cases the size and the concentration of stock market have an influence on poverty.While examining the impact of financial development on the poor, Beck, Demirgüç-Kunt, and Levine (2007) found that financial development disproportionately boosts the incomes of the poorest quintile and reduces income inequality.Honohan and Beck (2007) found that countries with deep financial systems seem to be associated with a lower incidence of poverty compared to those at the same level of national income.Jeanneney and Kpodar (2011) found that the poor benefit from the banking system's ability to facilitate transactions and provide opportunities for savings.Using data from 58 developing nations, Beck and Maimbo (2013) found that countries with better-developed financial intermediaries tend to have faster decreases in poverty.Uddin et al. (2014) found that financial development helps to reduce poverty in Bangladesh, although its effect is not linear.Abosedra, Shahbaz, and Nawaz (2016) found that financial development reduces poverty in Egypt when domestic credit to the private sector is used as a proxy for financial development.Bayar (2017) found that financial development, namely banking sector development and stock market development, have a significant positive impact on poverty reduction in emerging economies.Cepparulo, Cuestas, and Intartaglia (2017) found that financial development has a positive impact on poverty alleviation across all alternative measures of financial development and poverty in a sample of developing countries.While using data from a sample of developing countries over the period 2004-2015, Rewilak (2017) ) found that financial development may be poverty reducing.While examining the impact of financial development, economic growth and income inequality on poverty in India, Sehrawat and Giri (2018) found that financial development and economic growth help in poverty reduction.More recently, Appiah, Frowne, and Tetteh (2020) found that liquid liability and bank domestic credit have poverty reduction effects in five African emerging economies when poverty is measured by per capita household consumption.
Apart from the above studies that support a positive relationship between financial development and poverty reduction, in a few studies, it has been found that financial development does not benefit the poor proportionately.As an example, while examining the impact of financial development on poverty and inequality in Africa, Fowowe and Abidoye (2013) found that financial development has not had a significant effect on poverty in African countries.According to the study, this finding confirms the deficiencies in African financial systems.While examining, inter alia, whether financial development enhances the incomes of the poorest quintile, Rewilak (2013) found that financial development may alleviate poverty, but not universally.The author concluded that a 'one-size-fits-all' model does not work for different regions.Using panel data from 71 developing countries, Donou-Adonsou and Sylwester (2016) found that microfinance institutions (MFIs) do not have any impact on poverty regardless of the measure of poverty used.While assessing the effects of financial development on income inequality and poverty in emerging countries, Seven and Coskun (2016) found that neither banks, nor stock markets play a significant role in poverty reduction.Rashid and Intartaglia (2017), while examining the impact of financial development on poverty reduction in developing countries, found that financial development plays a significant role in reducing absolute poverty, but failed to find any pro-poor impact of financial development when poverty is measured in relative terms.Bolarinwa et al. (2021), while examining whether financial development really matters for poverty reduction in Africa, found that all financial development measures do not affect relative poverty.They found that only private credit reduces poverty in Africa, but only when the absolute measure of poverty is used.More recently, de Haan, Pleninger, and Sturm (2022), while examining whether financial development reduces the poverty gap in 84 countries, found that financial development does not have a direct effect on the poverty gap.

Methodology
The panel model used in this study to examine the longrun relationship between the various proxies of stock market development and poverty reduction in SSA countries can be expressed as follows: m it = Error term expected to be normally and identically distributed with zero mean and constant variance.Equation ( 1) can be estimated using various panel estimation techniques.In this study we used three panel data techniques, namely (i) the panel least squares; (ii) the fully modified ordinary least squares (FMOLS), 4 and (iii) the dynamic ordinary least squares (DOLS), to examine the impact of various stock market proxies on poverty reduction in the studied countries.The panel least squares (LS) method has been used in this study alongside the DOLS and FMOLS to check for robustness and to assess the sensitivity of our parameter estimates to different estimation techniques.

Heterogenous granger causality
In order to corroborate the results of the panel LS, DOLS and FMOLS, the heterogeneous panel Granger noncausality estimator based on Dumitrescu and Hurlin (2012) has been used.The advantage of using the heterogeneous panel Granger non-causality technique is that it takes into consideration the cross-section dependence in the data.In addition, it has been found to account for the time dimension, as well as the size of cross-section units relative to each other.The Dumitrescu and Hurlin (D-H) panel Granger non-causality model can be expressed as follows: where: y and x = variables, i.e. two stationary variables observed for N individuals on T periods; T = time dimension, i.e.
According to D-H, the null hypothesis of Homogeneous Non-Causality (HNC) means that there is no causal relationship for any of the cross-section units (Dumitrescu and Hurlin 2012).
The data used in this study span from 1993 to 2019.The data were obtained from the World Bank's Development Indicators (https://databank.worldbank.org/source/world-development-indicators) and the latest online version of the World Bank's Global Financial Development Database.The World Bank data were supplemented by individual countries' national statistical databases/reports.
Unfortunately, owing to lack of adequate data on stock market for many SSA countries, the study could not include data from all SSA countries.This data challenge has been exacerbated by the fact that the stock market in many sub-Saharan African countries is still in its infancy.Many countries' financial systems are still dominated by bank-based financial development sector.Consequently, only six middle-income countries, which were confirmed to have adequate and reliable data during the study period have been included in the current study.These include Kenya, Cote d'Ivoire, Nigeria, Ghana, South Africa, and Namibia.
In this study, the household consumption expenditure per capita has been used a proxy for poverty reduction in the studied countries.Indeed, the household consumption expenditure has been found to be one of the most reliable proxies for measuring the level of poverty by many studies.Some of the recent studies that have adopted household consumption expenditure as a proxy for poverty reduction include Olaniyi et al. (2022), Solarin, Gil-Alana, and Gonzalez-Blanch (2021), Musakwa and Odhiambo (2021), Dada and Akinlo (2021), Appiah-Otoo and Song (2021), Adeleye et al. (2020), Ho and Iyke (2018), Garza-Rodriguez (2018), Sehrawat and Giri (2016), Uddin et al. (2014), Odhiambo (2009); and Quartey 2005, just to mention but a few.
Although several proxies for measuring poverty have been proposed in the literature, some of them have been found to have limitations.For example, some studies have used annual income per capita as a proxy for poverty.Yet, annual income per capita does not account for other dimensions of poverty (see Odhiambo 2009).Moreover, previous studies have found that consumption expenditure among the poor communities is usually more reliably reported than income (see Woolard and Leibbrandt 1999;Ravallion 1992). 5Others have also argued that consumption is a better measure of welfare because it provides a more reliable information on standard of living of households than income (see Olaniyi et al. 2022;Koomson, Villano, and Hadley 2020;Beegle et al. 2012;World Bank 2001).It is also worth mentioning that the household consumption expenditure per capita used in this study is consistent with the World Bank's definition of poverty, which regards poverty as 'the inability to attain a minimal standard of living' measured in terms of basic consumption needs (see World Bank 1990).

Cross-sectional dependence (CD) and unit root tests
In accordance with previous studies, it is important to conduct a panel cross-sectional dependence test before conducting a panel unit root test. 6This will enable us to decide which unit root tests to employ in the study.For this purpose, four cross-sectional dependence tests have been used, namely: (1) cross-sectional dependence tests based on Breusch and Pagan (1980) LM; Pesaran (2004) scaled LM; Baltagi, Feng, and Kao (2012) bias-corrected scaled LM; and Pesaran (2004) CD.The results of cross-section dependence tests reported in Table 1 (Panel A) show that, on the whole, there is cross-section dependence in the data used.Consequently, the study used both the firstand secondgeneration unit root tests to test for stationarity in the series.The results of the unit root tests are reported in Panel B of Table 1.Based on these results, we can conclude that all the variables are integrated of order 1 [i.e.I (0)] and not of order zero [i.e.I ( 0 )].This has been confirmed by the first-generation unit root tests (LLC t-Statistics and IPS) and the second-generation unit root test (Pesaran -CIPS).

Panel cointegration test
Having found that all the variables are integrated of order one, our next step is to test whether the variables are cointegrated.For this purpose, two cointegration tests have been used, namely the Pedroni (2004) residual cointegration test 7 and the Westerlund (2007) cointegration test.The results of the Pedroni (2004) and Westerlund (2007) cointegration tests are reported in Tables 2  and 3, respectively.
The results reported in Table 2 show that all the variables included in Models 1, 2 and 3 are cointegrated based on the Pedroni residual-based cointegration test.This has been confirmed by the Panel PP-Statistic, Panel ADF-Statistic, Group PP-Statistic and Group ADFstatistic, which have been found to be statistically significant in Models 1, 2 and 3.The results of the Pedroni test have also corroborated by the Westerlund cointegration results reported in Table 3.This has been confirmed by the Westerlund Robust p-values of Gt, Pt, and Pa, which have been found to be statistically significant in Models 1, 2 and 3.These results show that even in the presence of cross-section dependency, the variables included in Models 1, 2 and 3 are still cointegrated.

Panel least squares, fully modified OLS (FMOLS) and dynamic OLS (DOLS)
Having confirmed that all the variables included in Models 1, 2, and 3 are all cointegrated, the next step is to examine the impact of the various proxies of stock market development on poverty reduction using the panel least squares, the panel DOLS, and the panel FMOLS 8 estimators.The empirical results based on these estimates are reported in Table 4.
The results reported in Table 4 show that the stock market has no significant impact on poverty reduction in the selected SSA countries during the period under study.This is confirmed by the coefficients of the stock market capitalization in Model 1, stock market traded value in Model 2 and stock market turnover ratio in Model 3, which have been found to be statistically insignificant.These results are supported by all the estimators used in this study, namely panel least squares, FMOLS, and DOLS estimators.These findings, though contrary to the results of some previous studies, are consistent with some of the previous studies, such as Seven   2016) for the case of emerging countries, Fowowe and Abidoye (2013) for case of African countries, Charlton (2008) for the case of developing countries when stock market liquidity is used as a proxy for stock market development, among others.In other results, it is shown that inflation has a negative impact on poverty reduction, as expected.This is confirmed by the coefficient of inflation in the poverty reduction equation, which has been found to be negative and statistically significant in all the three models.This finding applies, irrespective of whether the estimation is conducted using the panel least squares, FMOLS or DOLS.
The results also show that urbanization and labor have a positive impact on poverty reduction.This is supported by the coefficients of urbanization and labor in the poverty reduction equation, which has been found to be positive and statistically significant in Models 1, 2 and 3.As in the case of inflation, these results are robust, irrespective of whether the estimation is conducted using the OLS, DOLS and FMOLS.Contrary to the expectations, the results show that economic growth has no significant impact on poverty reduction in the countries under study.This also applies, irrespective of the model estimated and the estimation technique used.This finding, though contrary to our expectation, is consistent with some previous studies, in which it has been found that economic growth may not trickle down to the poor and the poorest if it is not pro-poor and inclusive.In some instances, it has been found even to trickle up to the rich or the middle class, thereby locking the poor in a vicious cycle of poverty.

Heterogeneous panel causality analysis
In this study, the Dumitrescu and Hurlin (2012) panel Granger-causality model was used to examine the causal relationship between the three various proxies of stock market development and poverty reduction.As the Dumitrescu and Hurlin (D-H) approach requires the variables to be stationary, the series had to be converted into the first difference before conducting a causality test.The results of the causality between the various proxies of stock market and economic growth are reported in Table 5.
The empirical results reported in Table 5 show that stock market does not Granger-cause poverty reduction in the countries under study, irrespective of the proxy used to measure the level of stock market.However, poverty reduction variable is found to Granger-cause stock market development when stock market development is proxied by stock market turnover.These results are supported by the respective Zbar-Statistics in the poverty equations in Models 1, 2 and 3, which have been found to be statistically insignificant and the Zbar-Statistic in Model 3, which have been found to be statistically significant when stock market is proxied by stock market turnover ratio.These results, therefore, corroborate the results of the panel least squares, panel DOLS and panel FMOLS, which also found that stock market has no significant impact on poverty reduction when these three proxies were used to measure the level of stock market development.Other results show that when stock market capitalization is used a proxy for stock market development (Model 1), (i) household Finally, when stock turnover ratio is used as a proxy (Model 3), (i) inflation Granger-causes poverty reduction, and (ii) inflation Granger-causes stock market development.

Conclusion
This study examined the relationship between stock market development and poverty reduction in the selected SSA countries during the period 1993-2019.notion that financial development in general is pro-poor, at least in the case of stock market development.
findings also highlight some of the deficiencies in SSA financial systems.Moreover, the prima facie evidence, which shows that poverty reduction Granger-causes the stock market turnover, is an indication that the development of the stock market in SSA countries is, in part, hindered by the poverty that is currently ravaging many SSA countries.Indeed, with the underdeveloped nature of capital markets in many SSA countries, it is highly unlikely that an increase in stock market development can reduce poverty in the studied countries significantly.
Besides, previous studies have also shown that financial development in general is unlikely to reduce poverty significantly if the level of income inequality is high since the benefits that accrue from the financial markets will simply trickle up to the rich and the upper-middle class rather than trickle down to the poorest segment of the society.The study, therefore, cautions SSA countries against over-relying on hot money, which results from stock market activities to reduce their poverty level.Instead, they should focus largely on policies that are aimed at alleviating poverty, which currently seems to act as an impediment to stock market development.In addition, they should focus on the preconditions that are necessary for the stock market to thrive.These include, amongst others, macroeconomic stability, a well-developed financial system, and a strong institutional and regulatory DEVELOPMENT STUDIES RESEARCH framework.In particular, a stable macroeconomic environment in the form of high economic growth, low inflation fiscal balance, for example, has been found to be a catalyst for foreign capital inflows, which are necessary for the development of the capital markets.A strong institutional and regulatory framework is also necessary in order to correct any mispricing in the financial market and to discourage speculators from exploiting asset price misalignments, thereby ensuring that financial markets are strong and stable. Notes ) where i = 1, . .., N represents the cross-sectional observation; t = 1, . .., T refers to the time period; Povertyred = Poverty reduction (proxied by real household consumption expenditure per capita); Stock = Stock market developmentproxied by (1) Stock market capitalization (Stkcap/GDP) -Model 1; (2) Stock market traded value (Stktrade/GDP) -Model 2; and (3) Stock market turnover ratio (Turnover/GDP) -Model 3; INF = Inflation; Urban = urbanization; y = GDP per capita; Labor = Labor force; d it and b 1i = Country specific effects and deterministic trend effects, respectively;