On the Nonlinear Effects of Globalization on Poverty: Insights from African Countries

Abstract

The study examines both the linear and nonlinear effects of globalization in 38 African countries between 1996 and 2018. In doing this, the study employs both linear and nonlinear autoregressive distributed lag estimation techniques. The findings from the two approaches show that globalization reduces poverty in Africa. Specifically, the findings from the symmetric models imply that the selected countries have benefited, in terms of poverty reduction, from an increase in trade and information flows occasioned by the current wave of globalization. On the other hand, the results from the asymmetric effect of globalization on poverty suggest that positive and negative shocks in overall and social globalization ameliorate poverty in the selected African countries. Thus, the study recommends that African governments should embark on those activities that will help in increasing the level of globalization for poverty to reduce in the region.

Keywords

Globalization poverty nonlinear ARDL Africa

Introduction

Globalization is one of the global phenomena experienced in the world. The reason for this is seen in the increased use of the internet, lower transportation costs, low tariffs, low restrictions on trade, increased number of treaties signed and embassies, the spread of ideas and information, remittances, and increased flows of capital. According to Dollar (2004), technological changes in transport and communication in addition to deliberate policy implementations are reasons for the rapid rise in integration among countries. While Deyshappria (2018) acknowledges the advent of the internet and telecommunications as reasons for rapid globalization, a bigger reason is the fall of communism and the wide acceptance of laissez-faire and free-market economies. Based on these, many countries in the world, African countries inclusively, are gearing efforts toward integrating with the world. For instance, most African countries have improved tremendously when it comes to trade liberalization according to Tsikata (2001). Ghana, Uganda, and Zambia are strong liberalizers, while Nigeria and Zimbabwe are weak liberalizers. Cote d’Ivoire, Kenya, Mauritius, South Africa, and Tanzania are found between strong and weak liberalizers and they are called intermediate liberalizers. Similarly, this categorization tallies with Brahmbhatt’s and Dadush’s (1996) integration index where the majority of the African countries are weak and slow liberalizers, except for Ghana and Mauritius which are strong liberalizers.

Meanwhile, globalization itself encompasses different aspects which include economic, social, and political according to Dreher (2006). As a result, Dreher (2006) through these dimensions of globalization, constructed the KOF globalization index and this is mostly used by different authors. The index was improved by Dreher, Gaston, and Martens (2008) and recently by Gygli, Haelg, Potrafke, and Sturm (2019). Globalization through these aspects is found to be affecting people’s standard of living. Since poverty is severe in most African countries, then integrating with the rest of the world would play a greater role in reducing the countries’ poverty levels. In line with this, Nissanke and Thorbecke (2008) postulate that globalization is a free force for poverty reduction since it helps in achieving faster economic growth and transmission of knowledge through the creation of a conducive environment. Globalization seemingly creates a world without borders, in which all countries can plug into a worldwide economy. However, the reverse is the case in African countries as many people wallow in absolute poverty with the improved level of globalization. To the authors, the reason for this variant is seen in the structural factors and policies implemented in these African countries. Some scholars believe that globalization is a subtle form of imperialism, crafted by the industrialized countries to keep African nations in their state of underdevelopment (Modebadze, 2012; Uzonwanne, 2018). L’Huille (2016) sees equality and reduction in poverty as a result of globalization to be a pipe dream, something unachievable given the presence of a trade model largely favorably skewed toward the industrialized economies. Despite these pessimistic views on the effects globalization would have on poverty in a country, some scholars are of the opinion that globalization, when paired with certain complementary reforms, would lead to better economic fortunes for less developed countries (LDCs). Dollar (2004) and Harrison and McMillan (2006) point to complementary policy options such as higher trade openness, the presence of social safety nets, food aid, and higher government expenditure as examples.

Statistically, Kharas, Hamel, and Hofer (2018) project that 70 percent of the world’s poor will be living in Africa in 2019 and this figure would have risen to 80 percent in 2023. Figure 1 displays the level of globalization (using the overall globalization index [OGI]) in the top 10 poorest countries in Africa and the world in 2018 according to Giovetti (2019). This shows that the prevalence of poverty is high in Africa when compared to the other regions. It is then obvious from the figure that the top 10 poorest countries in Africa still witnessed a slightly improved level of globalization. The question that comes to mind is why these countries experience a high level of poverty in the face of globalization.

Figure 1.

Source: KOF Globalization Index (Dreher, 2006; Dreher et al., 2018; Gygli et al., 2019).

In an empirical way, several earlier studies have shown that the effect of globalization on poverty is mixed. While some studies (Agenor, 2004; Bergh & Nilsson, 2011, 2014; Goff & Singh, 2014; Gohou & Soumare, 2012; Majeed & Farooq, 2021; Ogbuaku, Adebisi, & Feridun, 2006; Oyewale & Amusat, 2013; Salahuddin, Vink, Ralph, & Gow, 2019; Salimono, 1999; Ucal, 2014) argue that the poor would benefit from the process of globalization, others (Agenor, 2004; Kanbur, 2014; Ogbuaku et al., 2006; Okungbowa & Eburayolo, 2014; Stark, 2004) believe that globalization process would increase the level of poverty.

From the discourse, the effect of globalization on poverty is slippery, hard to pin down, and thus has no definite result on a given country because it follows a random path. In addition, there are multiple ways, it could affect the populace of a country (Sindzingre, 2005). Modebadze (2012) compares the effect of globalization on poverty to both sides of a coin, as it can have either positive or negative effects, based on the region. Just as globalization has the potential to bring about higher living standards, it can also lead to the creation of poverty traps for the population on the lowest rungs of the income ladder. In as much as the discourse has raged on, it is obvious that there are several factors at play in determining the effect of globalization on a country. Sindzingre (2005) believes this is majorly due to the presence and level of development of social institutions, which play a key role in the impact of globalization on poverty. This ensures that globalization has various effects across different countries. In the view of Harrison and McMillan (2006), there is little in the way of direct linkages between globalization and poverty, making the two concepts hard to study with a definite result for each country.

It is on this note that this current study intends to examine the effect of globalization and its dimensions on poverty in some selected African countries. The study employs different dimensions of globalization with the overall globalization because the various dimensions could impact poverty differently as most studies focus primarily on the economic aspect of globalization. Also, economic globalization as one of the dimensions of globalization is widely accepted as a measure of globalization while other dimensions are neglected in reducing poverty (Ajide, Raheem, & Asongu, 2018). Social globalization through the use of the internet, remittances, spread of ideas, information and images can help in reducing poverty as people use the internet to be involved in economic activities that will yield income, whereas remittances flow can also be used for any economic activities, thus reducing the poverty level. In addition, political globalization as a result of the number of embassies in the host countries and the number of treaties signed can lead to employment generation that will help in reducing poverty.

Apart from the above, this study wants to investigate the nonlinear (asymmetric) effect of globalization on poverty. This is premised on the fact that globalization could have a nonlinearity effect on the poverty level as against its linear effect on poverty as discussed by other studies. According to Koengkan, Fuinhas, and Santiago (2019), the process of globalization could have both negative and positive impacts on the economies. For instance, Osinubi (2020), Lee (2014), and Kutor (2014) have shown that globalization helps in reducing poverty, while studies such as Majeed and Farooq (2021), Osinubi (2020), Deyshappria (2018), Kanbur (2014), Okungbowa and Eburajolo (2014), and Singh and Huang (2011) document that globalization has done more harm than good in alleviating poverty. To add to this, Chishti, Ullah, Ozturk, and Usman (2020) report that the asymmetric effects of variables seem more practical, as the behavior of people (variables) cannot be predicted with certainty. In the same study, the authors argue that asymmetric modeling has more power to reveal more detailed and reliable findings (Katrakilidis & Trachanas 2012; Lee & Huang, 2017; Li, Li, & Si, 2019; Marques, Fuinhas, & Tomás, 2019). Thus, a need to investigate this effect arises. The asymmetric effect would enable us to see the positive and negative effects of globalization on poverty because both effects might not produce the same outcome.

The organization of this study is given as follows. Following this introductory section is the “Literature Review” section where previous studies on the subject matter are discussed. The “Data Sources and Methodology” section details the methodology used in achieving the study objective. Empirical results with the discussion of findings are presented in the “Empirical Results and Discussion” section, while the “Concluding Remarks” section gives the concluding remarks with some recommendations for African governments.

Literature Review

There are ample studies on the relationship between globalization and poverty in both developed and developing countries. Also, the studies use different dimensions of globalization such as trade openness, financial openness, technological openness, and foreign direct investment, and all of these are categorized under economic globalization. Studies that employ all three dimensions—economic, social, and political globalization—are few. One such is the work of Khan and Mujeed (2018) which aims to explore the impact of the various dimensions of globalization on cross-country poverty. Their findings show that both economic and social globalization reduce poverty, while political globalization does not. The studies that establish a negative relationship between globalization irrespective of its measure are discussed as follows. Stark (2004) argues that globalization leads to the improvement of human capital in poor countries and this would cause the level of poverty to diminish in the long run. In line with this, Agenor’s (2004) study supports Stark (2004) by establishing that globalization causes improved economic growth that would help in alleviating poverty in the long run. In studying 26 developing countries, Ucal (2014) reveals that globalization (FDI) negatively affects poverty. In the same spirit, Bergh and Nilsson (2011, 2014) study the relationship between poverty and globalization in 114 countries and their findings show that there is a negative connection between globalization and poverty in these countries. In emerging economies, it is evident from the study of Osinubi (2020) that economic globalization increases the level of poverty in Indonesia and Turkey, while it reduces the poverty level in Mexico. In addition to this, the poverty level in Mexico and Turkey increases, while it decreases in Indonesia as a result of a higher level of social globalization. The study also reports that political globalization results in a low level of poverty in Indonesia and Turkey, while its impact on the poverty level is insignificant in Mexico. In Pakistan, Majeed and Farooq (2021) establish that there is a negative relationship between globalization and annual poverty in the said country. This implies that as the level of globalization increases, annual poverty in Pakistan reduces.

Furthermore, the studies on the subject matter in African countries are few. Koffi, Gahe, and Ping (2018) in sub-Saharan Africa, using the Foster-Greer-Thorbecke (FGT) measure of poverty, reveal that globalization through international trade reduces poverty in these countries, ceteris paribus. In addition, Round (2007) confirms that the process of globalization is slow in African countries and this adversely affects the growth rate and increases the level of poverty. In like manner, Modebadze (2012) draws double-edged conclusions from the study of the positive and negative sides of globalization. Though the Asian Tigers have enjoyed huge success from globalization, experiencing higher living standards and greater prosperity, the sub-Saharan nations have been plunged into impoverishment and economic crises. Similarly, Deyshappria (2018) attempts to examine not only the general effects but also the specific effects of globalization on poverty and discovers that generally, globalization, in tandem with the secondary school enrollment ratio; percentage of urban population, and percentage of people with access to electricity all reduce poverty.

However, in terms of region-specific effects, the poverty-reducing attribute of globalization is strongest in South Asia, followed by East Asia, Europe, then Central Asia, with sub-Saharan Africa last on the list. Nissanke (2009) is not too far off with his view, opining that globalization has made the poor more vulnerable through macroeconomic shocks and lowered their bargaining power in global value chains. Using a panel of 30 African countries, Goff and Singh (2014) report that trade openness would help in reducing poverty in those countries with a financial sector deepening, a higher level of education, and stronger institutions. In testing the direction of causality between globalization (FDI) and poverty in five African regions, Gohou and Soumare (2012) show that FDI significantly and positively affects poverty in Central and East Africa, while the effect is insignificant in Northern and Southern Africa. However, an ambiguous effect is found between FDI and poverty in West Africa. The reason for the deviations in these results could be due to different economic policies put in place in each of the regions. L’Huiller (2016) is largely opposed to the idea of globalization lifting sub-Saharan countries out of the clutches of poverty, believing the system is set up for African countries to fail. According to her, the business model of globalization is largely unfavorable to sub-Saharan countries, as the globalization business model has stripped LDCs of trade protection policies, while the more developed countries (MDCs) retain their protectionist policies. In addition, the utilization of resources of LDCs is done by international markets, with little regard for citizens of the home countries.

Specifically, studies on the Nigerian economy as one of the African countries show that globalization when measured by trade openness and domestic investment helps in reducing poverty (Ogbuaku et al., 2006; Okungbowa & Eburayolo, 2014), globalization increases poverty when it is measured by financial openness, and foreign direct investment, FDI, as a proxy of globalization, has an insignificant effect on poverty (Okungbowa & Eburayolo, 2014). However, Oyewale and Amusat (2013) opine that globalization could only reduce the level of poverty in Nigeria if only the poor are involved in the process of income-growth activities; otherwise, the process of globalization would do more harm than good for the poor. In a related study of unbundling the process of globalization, Osinubi (2020) reveals that economic and social globalization adds to the level of poverty, while political globalization reduces the poverty level in Nigeria. Uzonwanne (2018) submits that globalization is a smokescreen designed to keep economic domination in the hands of the MDCs through economic programs. Her study uses chi-square and F-distribution to study the impact of economic globalization on African countries, with a specific focus on Nigeria. The results show that despite the subsequent rise in GDP as a result of globalization, income inequality, and poverty have worsened, as the fruits have been enjoyed by a low proportion of Nigerians. This is closely shadowed by Hassan (2013), who seeks to take a broader view of the situation, examining the effect of globalization on not just poverty, but the Nigerian economy as a whole. Using the presence of FDI as a means of globalization, OLS results show that while GDP, imports, and exports have risen due to the influx of FDI, for the vast populace, globalization has led to higher reliance on foreign capital, more marginalization from the world economy, and subsequently a deeper poverty level. A thread that can be drawn from the works of Uzonwanne (2018) and Hassan (2013) is the occurrence of economic growth as a result of globalization, yet devoid of improved living standards for the majority of Nigerians.

In South Africa, Salahuddin et al. (2019) find out that globalization alleviates poverty in the country. Finding a causal link between globalization and poverty, Magombeyi and Odhiambo (2017) investigate whether a causal link exists between globalization, using FDI, and poverty reduction, using life expectancy, infant mortality rate, and household consumption expenditure (HCE). The authors find that poverty reduction, when measured by life expectancy and infant mortality rate, causes FDI in both the short- and long-run and not otherwise, while there is no evidence of causality between FDI and poverty reduction using HCE.

It can be summarized that the relationship between globalization, irrespective of its measure, and poverty is inconclusive—a positive, negative, and insignificant relationship. Also, it is obvious from the review that globalization alone cannot help in reducing the poverty level. This implies globalization with an improved level of education, financial sector deepening, stronger institutions, improved economic growth, and involvement of the poor in income-growth activities among the poor to mention a few would cause the poverty level to reduce. The different findings could be due to the different methodology employed, countries studied, and measures of globalization used. Therefore, this current study deviates from the extant studies by studying the asymmetric effect of globalization on poverty and also by using all three dimensions of globalization instead of the conventional economic globalization (trade openness, financial openness, and FDI) used by most studies.

Data Sources and Methodology

The study employs annual data with an unbalanced panel from 38 African countries to unveil the symmetry and asymmetric effects of globalization on poverty among the selected countries. The sample period covers between 1996 and 2018 implying a total of 23 years based on data available on the key variables. Premised on the existing studies on the globalization–poverty nexus, the study identifies variables such as globalization, economic growth, population growth, agricultural value-added, and institutional quality as key determinants of poverty, especially in the African context (Deyshappria, 2018; Orija & Folawewo, 2021; Salahuddin et al., 2020). Therefore, the baseline empirical model to address the study’s objective is presented as

{P O V}_{i t} = β_{0} + β_{1} {G L B}_{i t} + β_{2} {G D P}_{i t} + β_{3} {A G R}_{i t} + β_{4} {P O P}_{i t} + β_{5} {I N S}_{i t} + ε_{i t}

(1)

where subscript $i$ represents the cross-section units ( $i = 1, 2, 3, \dots, 38)$ and $t$ is the time period (1996–2018) and $ε_{i t}$ is the error term.

${P O V}_{i t}$ stands as the measure of poverty which can be HCE or life expectancy rate (LER). The two measures of poverty have been extensively adopted as indicators of poverty in the literature. HCE is used based on the argument that consumption is preferable to income because the former directly determines the quality of life of people (see Magombeyi & Odhiambo, 2017; Musakwa & Odhiambo, 2019; Stoyanova & Tokin, 2018). In the same way, life expectancy at birth serves as a comprehensive measure of health poverty. Health is a major component in the computation of the human development index and, thus, serves as a good indicator of poverty, especially in Africa where the LER is perceived to be low (see Orija & Folawewo, 2021).

${G L B}_{i t}$ captures the measures of globalization. However, for robustness’s sake, we use four indicators to proxy the globalization index, namely, the OGI, economic globalization index (EGI), political globalization index (PGI), and social globalization index (SGI). Each of the measures of poverty variables is regressed on the four globalization variables to isolate their individual effects on poverty among the selected African countries. A similar approach is employed by Ojeyinka and Osinubi (2022) and Osinubi (2020).

${G D P}_{i t}$ represents economic growth proxy with the per capita GDP of the selected countries at 2015 constant United States dollars. The study incorporates economic growth into the model to capture the impact of economic activity which has been widely identified as a key determinant of poverty (Deyshappria, 2018; Osinubi, 2020). ${P O P}_{i t}$ is population growth. Studies (Osinubi, 2020) have linked an increase in the poverty rate, especially in the African region, to rapid population growth in the continent, hence its incorporation into the model. ${I N S T}_{i t}$ is a measure of institutional quality sourced from the World Bank’s world governance indicators (WGI) namely voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption. Originally, the WGI ranged between −2.5 and 2.5 indicating weak and strong institutional quality, respectively. However, for a meaningful explanation of the coefficient of the institutional variable, the study rescales each of the governance indicators (using 0.0–5.0) by adding +2.5 to the actual index. For instance, a country with −2.5 (weak institution) on the WGI will correspond to 0.0 on the modified scale, while a country with 2.5 on WGI corresponds to 5.0 on the new scale. Hence, following this method, 0.0 and 5.0 represent weak and strong institutional quality, respectively. A similar approach is employed by Adedokun (2017), Ajide (2022), and Delavallade (2006). The study derives a single index to capture the institutional variable (INS) which is generated by averaging the values of the six dimensions of governance indicator. This approach has also been used to generate a single measure of institutional quality in the literature (see Le, Kim, & Lee, 2016; Peres et al., 2018). Lastly, on the control variable, the study also incorporates AGR which stands for agricultural value added (Deyshappria, 2018). This is based on the agrarian nature of the African economies. Importantly, all the variables, except for globalization (KOF Index) and institutional variables (WGI), are sourced from the World Bank’s world development indicators (WDI). All the variables are employed in their level form. The measurement and sources of the variables are presented in Table 1.

Table 1.

Measurement and Description of Variables.

Variables	Symbol	Measurement	Source
Household Consumption expenditure	HCE	Household and NPISHs final consumption expenditure per capita (constant 2015 US $)	WDI
Life expectancy at birth	LEX	Life expectancy at birth, total (years)	WDI
Overall globalization index	OGI	An index of overall globalization (0–100)	KOF index
Economic globalization index	EGI	An index of economic globalization (0–100)	KOF index
Political globalization index	PGI	An index of political globalization (0–100)	KOF index
Social globalization index	SGI	An index of social globalization (0–100)	KOF index
Economic growth	GDP	GDP per capita (constant 2015 US $)	WDI
Population growth	POP	Population growth (annual percent)	WDI
Agriculture value-added	AGR	Agriculture, forestry, and fishing, value-added per worker (constant 2015 US$)	WDI
Institutional Quality	INS	Constructed by the authors using the average of voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption (0–5)	WGI

Note: WDI indicates world development indicators.

Descriptive Statistics and Measurement of Variables

Before the actual estimation is performed, it is crucial to examine some salient attributes of the data in the study. The summary of the descriptive statistics of the variables in the study is provided in Table 2. Except for HCE, GDP, and AGR, all the variables show a high level of consistency as evident by the proximity of their mean and median. Besides, both mean and median values for all the series lie between their minimum and maximum values. Starting with the poverty indicator variables, the average HCE of the selected countries is $1,257 with a maximum value ($7,572) recorded in Mauritius (2018) and a minimum value ($192) in Mozambique. For the second proxy of poverty, the selected African countries record an average life expectancy at birth of approximately 58 years with 77 years and 35 years as maximum and minimum life expectancy recorded in Algeria and Rwanda, respectively. This again reflects the extent of health poverty in the continent when compared with other continents of the world. For globalization variables, the sampled countries achieve an average of 47.5, 44.6, 59.1, and 38.5 in the overall, economic, political, and social globalization, respectively, suggesting that countries in Africa are found to be low globalizers, except in political globalization where the continent is discovered to be moderately globalized (59 percent). The achievement in political globalization can be linked to some treaties signed by the concerned countries. Countries such as Mauritius are known to be the strongest globalizer with maximum values of 72.3 percent, about 85.2 percent and 78.5 percent in the overall, economic, and social globalization, while Egypt occupies the highest value (91.58 percent) in political globalization. On the other hand, Burundi is found to be the least globalizer in overall, economic and social globalization with 22.5 percent, 21.3 percent, and 10.4 percent, respectively, between 1996 and 2018 while Equatorial Guinea has the lowest value (20.9 percent) in political globalization among the selected countries. For the control variables, the average real GDP capita in the selected countries stands at $2,051 suggesting that most of the selected countries are in the lower income category of the World Bank classification. This again provides an answer to low mortality and higher poverty incidence in the continent. Again, the sampled countries are weak in institutional quality given an average indicator of 1.9 out of 5.0.

Table 2.

Descriptive Statistics.

	HCE	LEX	OGI	EGI	PGI	SGI	GDP	AGR	INS	POP
Mean	1,257.002	58.380	47.524	44.681	59.088	38.511	2,051.332	2,564.166	1.934	2.442
Median	842.554	57.568	47.009	44.892	59.751	37.539	1,177.250	1,564.850	1.911	2.612
Maximum	7,572.710	76.693	72.354	85.299	91.587	78.558	16,438.600	18,584.900	3.380	8.118
Minimum	192.651	35.380	22.532	21.252	20.851	10.396	236.461	228.993	0.400	−0.616
Std. Dev.	1,163.579	7.995	9.842	10.796	16.896	13.941	2,242.075	2,702.602	0.600	0.943
Observation	794	872	872	872	872	872	871	822	759	872

Estimation Strategy

The article employs the linear and nonlinear nonlinear autoregressive distributed lag (ARDL and NARDL) techniques propounded by Pesaran et al. (2001) and Shin et al. (2014), respectively, to estimate the models.

As a first step, the study presents the linear ARDL form of Equation (1) to capture both the short-run and long-run dynamics as follows:

Δ P O V_{i t} = φ_{0} + \sum_{d = 1}^{r} φ_{1 d} Δ P O V_{i t - d} + \sum_{d = 0}^{r} φ_{2 d} Δ G L B_{i t - d} + \sum_{d = 0}^{r} φ_{3 d} Δ G D P_{i t - d} + \sum_{d = 0}^{r} φ_{4 d} Δ A G R_{i t - d} + \sum_{d = 0}^{r} φ_{5 d} Δ I N S_{i t - d} + \sum_{d = 0}^{r} φ_{6 d} Δ P O P_{i t - d} + α_{1} P O V_{i t - 1} + α_{2} G L B_{i t - 1} + α_{3} G D P_{i t - 1} + α_{4} A G R_{i t - 1} + α_{5} I N S_{i t - 1} + α_{6} P O P_{i t - 1} + μ_{t}

(2)

From Equation (2), $and d$ represent the difference operator and lag, respectively, $φ_{i s}$ are the short-run coefficients, and $α_{i s},$ normalized by $α_{1},$ are the long-run parameters. In analysing Equation (2), the null hypothesis of no cointegration (FD ${: H}_{0} {: α}_{1} = α_{2} = α_{3} = α_{4} = α = α_{6} = 0$ ) is subjected to scrutiny against the alternate hypothesis of cointegration $H_{1} {: α}_{1} \neq α_{2} \neq α_{3} \neq α_{4} \neq α_{5} \neq α_{6} \neq 0)$ . To validate the existence of cointegration, the study compares the critical values provided by Pesaran et al. (2001). As a norm, the null hypothesis of no cointegration is rejected if the computed F-statistic exceeds Pesaran et al. (2001)’s upper critical bound, while there is an absence of cointegration lower critical bound that exceeds the estimated F statistic. The outcome is declared inconclusive if the value of the computed F test lies between the two bounds.

In the event of cointegration, the symmetric ARDL version of Equation (2) is presented as

Δ P O V_{i t} = δ_{0} + \sum_{d = 1}^{p} δ_{1 d} Δ P O V_{i t - d} + \sum_{d = 0}^{p} δ_{2 d} Δ G L B_{i t - d} + \sum_{d = 0}^{p} δ_{3 d} Δ G D P_{i t - d} + \sum_{d = 0}^{p} δ_{4 d} Δ A G R_{i t - d} + \sum_{d = 0}^{p} δ_{5 d} Δ I N S_{i t - d} + \sum_{d = 0}^{p} δ_{6 d} Δ P O P_{i t - d} + γ E C T_{i t - 1} + ε_{t}

(3)

From Equation (3), ${E C T}_{t - 1}$ captures the error correction term, $γ$ is the parameter associated with the ECT. The parameter ( $γ)$ explains the speed of adjustment with which any deviation from the short run converges to the established long-run equilibrium.

One major weakness of the linear ARDL model above is its failure to uncover the possibility of the nonlinear effect of globalization on poverty. Specifically, Equations (1)–(3) assume that the positive and negative changes in GLB on poverty are the same which is not realistic given the increasing wave of globalization in recent times. The key message here is that an increase and a decrease in GLB may impact poverty differently. This is the premise on which the notion of asymmetry is built. To capture the possibility of asymmetry in the globalization–poverty nexus, the study modifies Equation (2) by splitting the globalization variable (GLB) into its negative and positive components as provided below:

{G L B}_{i t} = {G L B}_{i t}^{+} + {G L B}_{i t}^{-}

(4)

where

G L B_{i t}^{+} = \sum_{i = 1}^{t} Δ G L B_{i}^{+} = \sum_{i = 1}^{t} \max (Δ G L B_{i}, 0)

(5)

G L B_{i t}^{-} = \sum_{i = 1}^{t} Δ G L B_{i}^{-} = \sum_{i = 1}^{t} \min (Δ G L B_{i}, 0)

(6)

Equations (5) and (6) present the partial sum of positive changes (increase) and negative changes in GLB, respectively.

Substituting ${G L B}_{i t}^{+} a n d {G L B}_{i t}^{-}$ for GLB in Equation (2), the asymmetry or nonlinear ARDL model in line with Shin et al. (2014) can be presented as

Δ P O V_{i t} = σ_{0} + \sum_{d = 1}^{n} σ_{1 d} Δ P O V_{i t - d} + \sum_{d = 0}^{n} σ_{2 d}^{+} Δ G L B_{i t - d}^{+} + \sum_{d = 0}^{n} σ_{2 d}^{-} Δ G L B_{i t - d}^{-} + \sum_{d = 0}^{n} σ_{3 d} Δ G D P_{i t - d} + \sum_{d = 0}^{n} σ_{4 d} Δ A G R_{i t - d} + \sum_{d = 0}^{n} σ_{5 d} Δ I N S_{i t - d} + \sum_{d = 0}^{n} σ_{6 d} P O P_{i t - d} + ω_{1} P O V_{i t - 1} + ω_{2}^{+} G L B_{i t - 1}^{+} + ω_{2}^{-} G L B_{i t - 1}^{-} + ω_{3} G D P_{i t - 1} + ω_{4} A G R_{i t - 1} + ω_{5} I N S_{i t - 1} + ω_{6} P O P_{i t - 1} + μ_{t}

(7)

Equation (7) captures the nonlinear ARDL model following the study of Shin et al. (2014). Here, the long-run positive and negative impacts of GLB on POV are represented by $ω_{2}^{+}$ and $ω_{2}^{-}$ , respectively. In the same, $σ_{2}^{+}$ and $σ_{2}^{-}$ are the effects of increase and decrease in GLB in the short run, respectively. Essentially, two steps are involved in estimating Equation (7). First is the probes into the possibility of asymmetry cointegration between POV and positive and negative components of GLB. The procedure is akin to that of the traditional linear ARDL where the computed F-statistic is compared with the lower and upper critical bounds. Again, to reject the null hypothesis that there is the absence of long-run asymmetry, the computed F-statistic must be greater than the upper bound provided in Pesaran et al. (2001).

Upon the establishment of long-run asymmetry, the error correction model associated with Equation (7) is presented below:

Δ P O V_{i t} = ρ_{0} + \sum_{d = 1}^{m} ρ_{1 d} Δ P O V_{i t - d} + \sum_{d = 0}^{m} ρ_{2 d}^{+} Δ G L B_{i t - d}^{+} + \sum_{d = 0}^{m} ρ_{2 d}^{-} Δ G L B_{i t - d}^{-} + \sum_{d = 0}^{m} ρ_{3 d} Δ G D P_{i t - d} + \sum_{d = 0}^{m} ρ_{4 d} Δ A G R_{i t - d} + \sum_{d = 0}^{m} ρ_{5 d} Δ I N S_{i t - d} + \sum_{d = 0}^{m} ρ_{6 d} Δ P O P_{i t - d} + \emptyset E C M_{t - 1} + μ_{t}

(8)

The second is to uncover the asymmetric effect of GLB on POV. To achieve this, the study computes the Wald test on the estimates of positive and negative components of GLB. This approach has been widely applied in asymmetry literature. The short-run asymmetry is confirmed if the magnitude of $Δ G L B_{t - 1}^{+}$ is significantly different from the coefficient of $Δ G L B_{t - 1}^{-}$ (i.e., $Σ_{d = 0}^{m} ρ_{2}^{+} \neq$ $Σ_{d = 0}^{m} ρ_{2}^{-}) .$

In the same way, the null hypothesis of long-run asymmetric is tested by computing the Wald test on the coefficient of ${G L B}_{i t - 1}^{+}$ and ${G L B}_{i t - 1}^{-}$ in Equation (7). Specifically, there is an indication of long-run asymmetry between FD on EFP in South Africa if: $- \frac{ω_{2}^{+}}{ω_{1}} \neq - \frac{ω_{2}^{-}}{ω_{1}}$ .

To estimate the ARDL model, the study adopts the pooled mean group (PMG) estimator propounded by Pesaran, Shin, and Smith (1999) based on its advantages over other panel estimation techniques. First, the PMG allows the simultaneous estimation of the short-run and long-run coefficients associated with the ARDL model. Second, the PMG combines both averaging and pooling of coefficients. Similarly, the estimator gives room for the long-run estimates to be homogenous across groups but allows the short-run parameters to vary across units. Thus, the PMG produces consistent estimates over other estimators such as mean group (MG) and dynamic fixed effects (DE).

Empirical Results and Discussion

Correlation Analysis

The study conducts correlation analysis on the regressors in the model. This is imperative to ensure that the models estimated are free from multicollinearity threats. The outcome of the correlation analysis is presented in Table 3. Except for the globalization variables, the levels of association among other explanatory variables are moderate with the highest correlation coefficient of 0.64 between INS and SGI. Since each measure of globalization enters different models, there is no threat from multicollinearity in all the model specifications.

Table 3.

Correlation Analysis.

	OGI	EGI	PGI	SGI	GDP	AGR	INS	POP
OGI	1.000
EGI	0.705	1.000
PGI	0.668	0.065	1.000
SGI	0.791	0.668	0.158	1.000
GDP	0.434	0.617	−0.087	0.579	1.000
AGR	0.486	0.334	0.168	0.587	0.487	1.000
INS	0.496	0.501	0.038	0.640	0.304	0.251	1.000
POP	−0.488	−0.491	−0.051	−0.603	−0.255	−0.520	−0.504	1.000

Cross-sectional Dependence Test

As a preliminary test, it is important to examine the possibility of cross-sectional dependence (CD) among the units in the panel. The concept of globalization has reduced the entire universe into a global village where countries depend on one another for political, economic, and social survival. Thus, it is unrealistic to assume that countries in the panel are independent of one another. To uncover the possibility of CD, we apply the methodology of Pesaran (2004) and subject each series in the study to the CD test. This approach has been widely employed in the literature (Olaniyi, 2022; Yıldırım, Gedikli, Erdoğan, & Yıldırım, 2020). Besides, the Pesaran (2004) test can be applied when the number of cross-sectional units (N) is greater than time (T) which is the case in the present study (Attard, 2019). The CD test is premised on the null hypothesis of cross-section independence among the units (countries) in the panel. The outcome of the CD test is contained in Table 4. The results in Table 4 reveal that the null hypothesis of cross-sectional independence is rejected for all variables except the indicator of institutional quality (INS). Thus, the outcome of the CD test validates the assertion that countries in the study are cross-sectionally dependent.

Table 4.

Pesaran CD Test.

Variable	CD Test	P value
HCE	73.934	.000
LEX	115.614	.000
GI	113.931	.000
EGI	17.905	.000
PGI	98.113	.000
SGI	122.764	.000
GDP	79.644	.000
AGR	26.822	.000
INS	0.697	.488
POP	3.089	.002

Panel Unit Root Tests

The presence of CD among the panel units implies that the use of first-generation panel unit root tests will produce inconsistent estimates due to the failure to account for CD in the series. Thus, the study employs the second-generation unit root test that accommodates CD to examine the stationary property of the variables in the study. Specifically, the study employs the Pesaran cross-sectional ADF (PCADF) test using the Stata command Pescadf proposed by Pesaran (2003). One major strength of the PCADF is that the test performs better with an unbalanced panel. The outcome of the panel unit root test is presented in Table 5. The results from the cross-sectional unit root test reveal that the series in the study are combinations of I(0) and I(1) variables. Specifically, the null hypothesis of unit root cannot be rejected for all the variables, except for life expectancy at birth (LEX), OGI, SGI, and population growth (POP) which are stationary at level. However, those series with unit root in their levels become stationarity at their first difference. The outcomes from the unit root test further provide support for the adoption of the ARDL since no variable is of higher order (I(2) and above).

Table 5.

PCDF Unit Root Test Results (Intercept and Trend).

Variables	Level	First diff.	Order
Variables	Z (t-bar) Test	Z (t-bar) Test	Order
HCE	−0.185	−6.387^***	I(1)
LEX	−21.317^***	–	I(0)
GI	−3.548^***	–	I(0)
EGI	1.277	−5.462	I(1)
PGI	−4.917	–	I(1)
SGI	−1.944^**	–	I(0)
GDP	4.022	−3.168^***	I(1)
AGR	0.543	−7.278	I(1)
INST	3.175	−3.167	I(1)
POP	−10.198^***	–	I(0)

Note: ^*** and ^** represent 1 percent and 5 percent significance level, respectively.

Symmetric Effects of Globalization on Poverty

The outcomes of the symmetric effect of globalization on poverty are presented in Table 6. Models 1, 2, 3, and 4 in Table 6 represent the specification for the OGI, EGI, PGI, and SGI, respectively, when household expenditure per capita (HCE) is used as an indicator of poverty. Meanwhile, Models 5, 6, 7, and 8 correspond to OGI, EGI, PGI, and SGI, respectively, using LER (LEX) as a poverty indicator. The long-run estimates are presented in Panel A, while Panel B contains the estimates from the short-run model. From the outcome in Panel B, the coefficients of the ECT(−1) assume the expected signs for all the specifications. The magnitudes of the ECT derived from the eight models are negative and significant. This indicates evidence of a cointegrating relationship between poverty indicators and other explanatory variables over the study period.

Table 6.

Symmetric Effects of Globalization on Poverty.

	Household Consumption Expenditure				Life Expectancy
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Panel A
Long-run results
GI	9.832^***				0.399^***
	(1.291)				(0.041)
EGI		7.723^***				−0.159^***
		(0.516)				(0.019)
PGI			−2.491^***				0.293^***
			(0.475)				(0.017)
SGI				5.147^***				0.245^***
				0.561				(0.013)
GDP	0.491^***	0.475^***	0.887^***	0.508^***	0.009^***	0.010^***	−0.0001	0.010^***
	(0.023)	(0.016)	(0.027)	(0.017)	(0.001)	(0.000)	(0.000)	(0.001)
AGR	0.059^***	0.052^***	−0.107^***	0.084^***	0.002^***	0.001^***	0.00045^***	0.0031^***
	(0.004)	(0.008)	(0.012)	(0.008)	(0.000)	(0.000)	(0.000)	(0.000)
INS	−64.960^***	−11.21	−18.67^*	−20.380^*	7.694^***	−4.669^***	7.317^***	4.135^***
	(22.490)	(8.048)	(11.160)	(11.310)	(0.736)	(0.441)	(0.523)	(0.469)
POP	−223.3^***	−24.24^*	−130.1^***	−11.895^**	26.70^***	10.887^***	12.339^***	15.701^**
	(28.750)	(14.010)	(8.012)	(5.258)	(1.266)	(0.567)	(0.374)	(0.629)
Panel B
Short-run results
D(GI)	−13.580^**				0.009
	(6.851)				(0.010)
D(EGI)		−3.690^**				0.007^**
		(1.491)				(0.003)
D(PGI)			−4.442				0.005
			(5.143)				(0.029)
D(SGI)				−3.753				0.008
				(3.647)				(0.012)
D(GDP)	0.207^***	0.275^***	−0.088	0.352^***	0.001	0.001	0.0021^**	0.001
	(0.078)	(0.096)	(0.148)	(0.083)	(0.001)	(0.001)	(0.001)	(0.001)
D(AGR)	−0.06	−0.129^**	0.002	−0.064^*	−0.001	−0.001^*	−0.0002	−0.001
	(0.039)	(0.056)	(0.047)	(0.038)	(0.001)	(0.000)	(0.001)	(0.001)
D(INS)	82.730^**	19.34	57.67	−36.038	−0.269^**	0.051	−0.312^*	−0.167
	(40.090)	(47.400)	(35.760)	(50.132)	(0.106)	(0.069)	(0.166)	(0.102)
D(POP)	306.700	59.560	−6.364	93.324	0.811^*	0.300	0.211	0.952
	(352.500)	(224.500)	(94.350)	(105.409)	(0.445)	(0.614)	(0.729)	(0.542)
ECM(−1)	−0.286^***	−0.298^***	−0.398^***	−0.186^***	−0.020^***	−0.034^***	−0.025^***	−0.027^**
	(0.061)	(0.065)	(0.073)	(0.044)	(0.007)	(0.012)	(0.008)	(0.010)

Note: Standard errors in parentheses; ^***p < .01, ^**p < .05, ^*p < .1.

Focusing on the long-run results in Panel A in Table 6, the outcomes of the long-run estimates show that the effect of GI (Model 1), EGI (Model 2), and SGI (Model 3) on HCE is positive and significant, suggesting that economic and social globalization reduce poverty in the selected African countries. Specifically, a unit increase in OGI, EGI, and SGI magnifies HCE by 9.8, 7.7, and 5.1 units, respectively. This implies that the effect of globalization on poverty is beneficial to the selected countries. On the other hand, the impact of PGI (Model 3) on HCE is detrimental in the long run as a unit increase in PGI reduces HCE by approximately 2.5 units. In the same vein, using life expectancy as a measure of poverty, all the dimensions of globalization (Models 5, 7, and 8) are found to enhance life expectancy in the selected countries with the exception of economic globalization (Model 6) where life expectancy decreases by 0.2 unit when economic globalization increases by 1 unit.

Overall, the outcome of symmetric ARDL suggests that globalization promotes poverty reduction among the sampled countries. The finding aligns with the “scale effect” of Cole (2006) that globalization boosts economic activity and by extension reduces poverty in the economy. The outcome also confirms the findings of Deyshappria (2018) for developing countries, Khan and Majeed (2018) for 113 developing countries, and Hassan, Bukhari, and Arshed (2019). To add to this, the finding validates the study of Salahuddin et al. (2019) for South Africa and Majeed and Farooq (2021) in Pakistan. All these authors confirm that globalization plays a pivotal role in poverty reduction in developing countries including Africa. The outcome shows that African countries have benefited from their integration into the world through globalization which has manifested in terms of poverty reduction in the continent.

For the control variables, the effect of GDP per capita (economic growth) on HCE and LEX is positive and significant, except for Model 7. The result implies that the economic growth of the selected countries is poverty-reducing in the long run. The outcome conforms with a priori expectation because an increase in economic activity is expected to stimulate employment opportunities which would, in turn, reduce the poverty level. The outcome is consistent under the two indicators of poverty. A similar finding is discovered by Khan and Majeed (2018), suggesting that, an increase in economic growth dovetails with better economic opportunities for the poor and vulnerable on the continent. Similarly, the effect of agricultural value-added on the two proxies of poverty is positive and significant, except for Model 3 when political globalization is employed. The finding is expected given the fact that the majority of the African population engages in agriculture-related activities. Thus, an increase in agricultural value-added would increase HCE, bring about an enhancement in life expectancy and consequently lead people out of poverty. The results imply that an increase in farmers’ productivity provides them with access to various resources including quality medical healthcare, which in turn elongates their life span. However, the impact of institutional quality and population growth on poverty is mixed and complex depending on the measure of poverty adopted. For instance, the coefficient of institutional quality is found to reduce HCE (see Models 1, 2, 3, and 4), and thus, worsens the poverty level, while its impact on the LER is positive and significant, which implies a reduction of poverty under Models 5, 6, and 8. Similarly, population growth exerts a negative and significant impact on HCE, while its effect on LEX is positive and significant indicating that the direction of the impact of POP on poverty depends on the indicator of poverty employed.

Interestingly, the outcomes from the short-run analysis in Panel B show that overall globalization (OGI) and economic globalization (EGI) significantly reduce poverty under Models 1 and 2, while the impact of other dimensions of globalization (PGI and SGI) is inconsequential when HCE is used to proxy poverty. In fact, only the impact of EGI (Model 6) is positive and significant, while the effects of other dimensions of globalization are not significant when poverty is measured by LER.

Again, the effects of economic growth on the two poverty indicators agree with the long-run estimates, though, with slight exemptions which overwhelmingly support the assertion that economic growth is a sine qua non for poverty reduction in the selected African countries. However, agricultural value-added is seen to aggravate poverty in the short run under the two proxies, while the impact of population growth is not significant for poverty reduction in the short run except for Model 5 when the overall globalization is regressed on LER at birth.

Asymmetric Effects of Globalization on Poverty

As argued in the introduction, there is increasing evidence that poverty may respond in a nonlinear manner (asymmetrically) to globalization due to the negative and positive spill-over effects arising from global interconnectivity among the economies of the world. However, despite the avalanche of studies on the globalization–poverty nexus, the issue of nonlinearity has not been considered, especially in the context of African countries. This study extends the scope of previous studies by incorporating the possibility of nonlinearity between globalization and poverty relationship using data selected from 38 African countries. To achieve this objective, we estimate the NARDL model specified in Equations (7) and (8), and the results of the long run and short un are reported in Tables 7 and 8, respectively.

The outcomes of the long-run NARDL models (Table 7) for all the control variables are similar to the findings from the linear ARDL reported in Table 6. For instance, economic growth is found to exert a positive and significant impact on HCE specifications (models 1, 2, 3, and 4), suggesting that an increase in economic activity enhances HCEs and by extension improves the welfare condition of the masses in the selected countries. However, the impact of economic growth on life expectancy is mixed in the long run (Models 5, 6, 7, and 8) conditioning the dimension of globalization considered. It is observed from Table 7 that economic growth (GDP) exerts a negative and significant impact on life expectancy at birth for models with overall globalization and SGI (Models 5 and 8) indicating that an improved economic growth worsens and accentuates the poverty rate in the selected countries. In contrast, the effect of economic growth on LEX is positive and significant in Model 5 where economic globalization is employed. This suggests that the selected countries benefit from the international flow of goods and services engendered by the globalization process.

Table 7.

Long-run Asymmetric Effects of Globalization on Poverty.

	Household Consumption Expenditure				Life Expectancy Rate
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Long-run results
OGI_Pos	5.278^***				1.331^***
	(1.074)				(0.051)
OGI_Neg	4.617^*				2.832^***
	(2.406)				(0.183)
EGI_Pos		−0.183				0.142^***
		(0.327)				(0.015)
EGI_Neg		−4.313^***				−0.223^***
		0.448				0.020
PGI_Pos			2.619^***				0.388^***
			(0.754)				(0.027)
PGI_Neg			−1.97^*				−0.145^***
			(1.062)				(0.041)
SGI_Pos				5.498^***				0.310^***
				(0.758)				(0.017)
SGI_Neg				18.190^***				0.185^**
				(5.310)				(0.097)
GDP	0.659^***	0.564^***	0.568^***	0.471^***	−0.002^***	0.004^***	0.004	−0.002^***
	(0.013)	(0.009)	(0.013)	(0.016)	(0.000)	(0.000)	(0.000)	(0.000)
AGRI	0.042^***	0.000	0.005	0.042^***	0.0001^**	0.000	0.000	0.005^***
	(0.005)	(0.002)	(0.005)	(0.007)	(0.000)	(0.000)	(0.000)	(0.000)
INS	−46.446^***	82.85^***	41.834^***	103.378^***	9.215^***	3.515^***	3.515^***	2.352^***
	(12.380)	(6.154)	(7.925)	(13.841)	(0.780)	(0.000)	(0.000)	(0.817)
POP	−5.766^***	−18.704^*	−20.347	−11.875^**	22.661^***	17.525^***	17.03^***	20.267^***
	(6.559)	(3.139)	(5.317)	(4.906)	(0.767)	(0.301)	(0.301)	(0.556)

Note: Standard errors in parentheses; ^***p < .01, ^**p < .05, ^*p < .1.

In the same spirit, the estimates of agricultural value-added are akin to the results of the symmetric ARDL in the long run as agricultural value-added is found to play a critical role in mitigating the incidence of poverty in these economies. Unlike the outcome from the symmetric model, the impact of institutional quality is found to significantly improve household consumption and elongate people’s lifespan in the selected countries, except in Model 1 where the impact of the OGI is poverty-aggravating. This suggests that a stronger institution provides a conducive atmosphere for people to engage in productive activities and thus enhance their well-being. Meanwhile, an increase in population growth constrains HCE in Models 1, 2, 3, and 4. The implication of this result is that rapid population growth drains existing resources per capita and thus deteriorates people’s welfare. However, the impact of population growth on LER is mixed depending on the dimension of globalization employed.

Focusing on the target variable, evidence from the long-run asymmetric effect of globalization on poverty (Table 8) connotes that the positive and negative shocks in overall globalization (OGI_Pos and OGI_Neg) have positive and significant effects on the two indicators of poverty in the selected countries (Model 1). In statistical terms, a unit increase in positive and negative shocks in OGI expands HCE by approximately 5.3 and 4.6 units, respectively, while a corresponding increase extends the LER by 1.3 and 2.8 units, respectively. This suggests that both the increase and decrease in the OGI are welfare-enhancing in the sampled countries, irrespective of the measure of poverty employed. However, in terms of magnitude, the coefficient of positive and negative shocks in OGI differs depending on the indicator of poverty adopted. For other dimensions of globalization, a positive shock (increase) in EGI (in Model 2) has no significant impact on HCE, while a negative shock in EGI produces a negative and significant effect on HCE in the long run. This further implies that any shock that reduces economic globalization in the sampled countries accentuates poverty. However, positive and negative shocks in EGI (Model 6) exert a positive and negative influence on LER, respectively. Hence, the study documents an asymmetric behavior between economic globalization and poverty indicators with negative shocks having a larger effect on poverty than the positive component. Interestingly, poverty indicators respond asymmetrically to political globalization (Models 3 and 7) where the positive component is found to have a stronger and larger effect on poverty. Specifically, a unit increase in positive and negative shocks to PGI increases and reduces the HCE in the long run by approximately 2.6 and 1.9 units, respectively. For LER specification (Model 7), a positive shock in PGI extends the LER by approximately 0.4 unit, while a corresponding negative shock (decrease) in PGI reduces LEX by approximately 0.1 unit. By implication, positive shock in political globalization ameliorates poverty, while a negative change accentuates poverty under the two poverty specifications. Lastly, both the positive and negative shocks in social globalization increase the HCE and LEX in the long run, although their effects vary depending on the indicator of poverty applied. Thus, the study establishes that the degree of asymmetry in the globalization–poverty nexus among the selected countries depends on the dimension of globalization and the indicator of poverty considered.

Table 8.

Short-run Asymmetric Effects of Globalization on Poverty.

	Household Consumption Expenditure				Life Expectancy
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
D(OGI_Pos)	7.634				0.052^**
	(0.515)				(0.020)
D(OGI_Neg)	3.169				−0.085
	(27.158)				(0.043)
D(EGI_Pos)		2.042				0.046^**
		(5.921)				(0.020)
D(EGI_Neg)		0.140				−0.043
		(6.238)				(0.031)
D(PGI_Pos)			2.833				0.003^**
			(9.495)				(0.001)
D(PGI_Neg)			20.35				(0.048)
			(54.536)
D(SGI_Pos)				3.602				0.062^**
				(3.448)				(0.031)
D(SGI_Neg)				−26.603				−26.603
				(35.785)				(35.785)
D(GDP)	0.322^***	0.349^***	0.451^***	0.384^***	0.000	0.002^***	0.003^**	0.384^***
	(0.108)	(0.107)	(0.087)	(0.084)	(0.000)	(0.001)	(0.001)	(0.084)
D(AGR)	−0.088	−0.085^**	−0.04	−0.021	−0.001	0.0003^**	−0.001^**	0.147^**
	(0.056)	(0.041)	(0.045)	(0.027)	(0.001)	(0.000)	(0.001)	(0.051)
D(INS)	5.715	23.224	−37.948	−34.744	−0.126	−0.105	−0.289^*	−0.326^**
	(55.268)	(38.405)	(59.662)	(46.450)	(0.260)	(0.177)	(0.194)	(0.165)
D(POP)	−326.009	−170.655	−62.644	16.485	1.352^***	0.559	−0.732	0.550
	(198.207)	(128.432)	(69.849)	(63.375)	(0.435)	(0.721)	(0.841)	(0.794)
ECM (−1)	−0.185^***	−0.239^***	−0.151^***	−0.190^***	−0.013^***	−0.020^***	−0.016	−0.021^***
	(0.044)	(0.000)	(0.037)	(0.047)	(0.003)	(0.001)	(0.013)	(0.047)
Wald Test for Asymmetry
Short run	4.728^**	3.275^**	1.260	4.194^**	6.169^***	0.112	5.503^**	48.218^***
Long run	2.100^*	1.749	1.612	0.505	1.331	0.692	0.047	0.262

Note: Standard errors in parentheses; ^***p < .01, ^**p < .05, ^*p < .1.

The short-run estimates of the NARDL on globalization and poverty nexus are further presented in Table 8. The estimates of the error correction terms are negative and significant for the estimated models, implying evidence of asymmetry cointegration among the variables of interest, except for Model 7 when political globalization is regressed on life expectancy. The outcomes from Table 8 show that both the positive and negative shocks in all the dimensions of globalization do not significantly impact HCE (Models 1, 2, 3, and 4) in the short run. In contrast, life expectancy at birth responds significantly to positive shocks in all the dimensions of globalization (Models 5, 6, 7, and 8). Again, the impacts of other control variables are similar to those reported under the long-run asymmetry in Table 7.

To further investigate the asymmetry relationship in the globalization–poverty nexus, the study subjects the coefficients of the positive and negative shocks in the four dimensions of globalization to the Wald test following the convention in the asymmetry literature. The findings of the Wald test are presented for each specification in the lower part of Table 8. Surprisingly, the results of the Wald test only support the existence of short-run asymmetry for all the specifications, except Model 6 which contains the EGI–LEX relationship, while the evidence of long-run asymmetry is only established for overall globalization (OGI) specification in Model 1. In general, one major finding emanating from the current study is the confirmation of asymmetry in the nexus between globalization and poverty. In the same way, the study has also unveiled the need to account for the effects of other dimensions of globalization on poverty to comprehensively capture their impact on poverty in the selected countries.

Concluding Remarks

The study pursues two principal objectives by examining the linear (symmetric) effect of overall globalization and its dimensions on poverty, as well as unveiling the existence of asymmetry behavior in the nexus between globalization and poverty in 38 selected African countries. Annual data between 1996 and 2018 are employed using both the traditional linear ARDL and NARDL as the techniques of estimation. Interestingly, we unbundle the globalization data to examine their differential effects on poverty in the selected countries. Thus, four dimensions of globalization are included in the study models in a step-by-step approach to avoid the problem of multicollinearity. For robustness, the study employs two proxies to measure poverty which are the HCE to capture expenditure poverty and life expectancy at birth to measure health poverty.

As a preliminary step, we establish the existence of CD among the panel units, and this informs us to employ the second-generation panel unit root test to unravel the order of integration of the series in the model. Based on the outcomes of the unit root test, the study applies the pool mean group estimator to estimate the NARDL models, based on its advantages over other estimators.

The outcome from the symmetric (linear) ARDL overwhelmingly supports the poverty-reducing effect of overall globalization and its dimensions for the two indicators of poverty utilized. It is also important to state that economic and social globalization account for much of the reduction in poverty experienced in the region. The findings from the symmetric models imply that the selected countries have benefited, in terms of poverty reduction, from an increase in trade and information flows occasioned by the current wave of globalization. On the other hand, the results from the asymmetric effect of globalization on poverty suggest that positive and negative shocks in overall and social globalization ameliorate poverty in the selected African countries. The finding is consistent under the two poverty indicators, and this suggests that welfare is improved as the selected countries become more socially globalized. Meanwhile, it is also discovered that positive (negative) shock in economic and political globalization reduces (accentuates) the poverty level in the selected countries. This suggests that African countries must further engage in activities that will pave way for the process of globalization in the region. On the control variables, the study finds that economic growth and agricultural value-added play significant roles in poverty reduction among the sampled countries while an increase in population growth is found to accentuate poverty level. The outcome is consistent under the symmetry and asymmetry specifications. Lastly, the results from the Wald test for the asymmetry support the existence of short-run asymmetry in all the estimated poverty models and that of the long-run asymmetry in the overall globalization and poverty using the HCE as a poverty indicator. The implication is that the issue of asymmetry must not be ignored in the discussion of the globalization–poverty nexus.

The recommendation emanating from the findings is that African governments should embark on those activities that will help in increasing the level of globalization for poverty to reduce in the region. The activities include the following: (a) lowering trade and investment barriers to reaping the gains of globalization, (b) consolidating the existing policies to improve agricultural productivity, and (c) improving information and communication technology to further strengthen trade relations between African countries and the rest of the world.

Future studies can examine the role of institution (either substitutability or complementarity) in the nonlinear effect of globalization on poverty as this present study focuses only on the direct effect of institution on poverty. This becomes imperative as institutional quality in Africa is weak. Also, there are limited studies to justify the findings of the asymmetric effects of globalization on poverty, thus, other studies can delve into this to add to the existing literature. On a final note, this study is constrained by data availability.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Attard

(2019). Public debt and economic growth nexus: A dynamic panel ARDL approach (MPRA Paper No. 96023).

Adedokun

J.A.

(2017). Foreign aid, governance and economic growth in sub-Saharan Africa: Does one cap fit all? African Development Review , 29(2), 184–196.

Agenor

P.R.

(2004). Does globalization hurt the poor? International Economics and Economic Policy , 1(1), 21–51.

Ajide

F.M.

(2022). Entrepreneurship and productivity in Africa: The role of institutions. Journal of Sustainable Finance & Investment , 12(1), 147–168.

Ajide

K.B.

, Raheem

I.D.

, & Asongu

S.A.

(2018). Dollarization and the “unbundling” of globalisation in sub-Saharan Africa. Research in International Business and Finance , 1–26. https://doi.org/10.1016/j.ribaf.2018.09.002

Bergh

, & Nilsson

(2011). Globalization and absolute poverty—A panel data study (IFN Working Paper No. 862). Research Institute of Industrial Economics.

Bergh

, & Nilsson

(2014). Is globalization reducing absolute poverty? World Development , 62(C), 42–61. https://dx-doi-org.web.bisu.edu.cn/10.1016/j.worlddev.2014.04.007

Brahmbhatt

, & Dadush

(1996). Disparities in global integration (Background Paper Prepared for Global Economic Prospects 1996). World Bank.

Chishti

M.Z.

, Ullah

, Ozturk

, & Usman

(2020). Examining the asymmetric effects of globalization and tourism on pollution emissions in South Asia. Environmental Science and Pollution Research , 27, 27721–27737. https://doi.org/10.1007/s11356-020-09057-9C

10.

Cole

M.A.

(2006). Does trade liberalization increase energy use? Economics Letters , 92, 108–112.

11.

Dollar

(2004). Globalization, poverty, and inequality since 1980 (Policy Research Working Paper 3333, pp. 1–46). World Bank

12.

Delavallade

(2006). Corruption and distribution of public spending in developing countries. Journal of Economics and Finance , 30, 222–239. https://doi.org/10.1007/BF02761488

13.

Deyshappria

(2018). Globalization–poverty nexuses: Evidences from cross-country analysis. Empirical Economic Review , 1(1), 25–48.

14.

Dreher

(2006). Does globalization affect growth? Evidence from a new index of globalization. Applied Economics , 38(10), 1091–1110.

15.

Dreher

, Gaston

, & Martens

(2008). Measuring Globalisation: Guaging its consequences (pp. 16–224). Springer. https://doi.org/10.1007/978-0-387-74069-0

16.

Giovetti

(2019). The world’s ten poorest countries . https://www.concern.net/news/worlds-ten-poorest-countries

17.

Goff

M. L.

, & Singh

R. J.

(2014). Does trade reduce poverty? A view from Africa. Journal of African Trade , 1(1), 5–14. https://dx-doi-org.web.bisu.edu.cn/10.1016/j.joat.2014.06.001

18.

Gohou

, & Soumare

(2012). Does foreign direct investment reduce poverty in Africa and are there regional differences? World Development , 40(1), 75–95.

19.

Gygli

, Haelg

, Potrafke

, & Sturm

J-E.

(2019). The KOF globalization index—Revisited. Review of International Organizations , 14(3), 543–574. https://doi.org/10.1007/s11558-019-09344-2

20.

Harrison

, & McMillan

(2007). On the links between globalization and poverty. The Journal of Economic Inequality , 5, 123–134. https://doi.org/10.1007/s10888-006-9041-9

21.

Hassan

O.M.

(2013). An appraisal of the effects of globalization on the Nigerian economy. American Journal of Business and Management , 2(4), 296–303.

22.

Hassan

M.S.

, Bukhari

, & Arshed

(2019). Competitiveness, governance and globalization: What matters for poverty alleviation? Environment, Development and Sustainability , 22(4), 3491–3518. https://doi.org/10.1007/s10668-019-00355-y

23.

Kanbur

(2014). Globalization and inequality (Working Paper No. 2014-03). Charles H. Dyson School of Applied Economics and Management, Cornell University.

24.

Katrakilidis

, & Trachanas

(2012). What drives housing price dynamics in Greece: New evidence from asymmetric ARDL cointegration. Economic Modelling , 29(4), 1064–1069.

25.

Khan

F.N.

, & Majeed

M.T.

(2018). Globalization and poverty nexus: A panel data analysis. Forman Journal of Economic Studies , 14, 143–174.

26.

Kharas

, Hamel

, & Hofer

(2018). Future development: Rethinking global poverty reduction in 2019 . Brookings Institutions. Retrieved, 9th June 2020, from https://www.brookings.edu/blog/future-development/2018/12/13/rethinking-global-poverty-reduction-in-2019/#:~:text=In%202019%2C%20some%2070%20percent,from%2060%20percent%20in%202016

27.

Koffi

S.L.Y.

, Gahe

Z.S.Y.

, & Ping

Z.X.

(2018). Globalization effects on sub-Saharan Africa: The impact of international trade on poverty and inequality. International Journal of Innovation and Economic Development , 4(3), 41–48. https://doi.org/10.18775/ijied.1849-7551-7020.2015.43.2005

28.

Koengkan

, Fuinhas

J.A.

, & Santiago

(2019). Asymmetric impacts of globalisation on CO ₂ emissions of countries in Latin America and the Caribbean. Environment Systems and Decisions , 1–13. https://doi.org/10.1007/s10669-019-09752-0

29.

Kutor

S.K.

(2014). Processes of globalization and its effects on the life of people in the global south by drawing examples from Ghana. Research on Humanities and Social Sciences , 4(6), 95–99.

30.

Lee

K.K.

(2014). Globalization, income inequality and poverty: Theory and empirics. Social System Studies , 28, 109–134.

31.

Lee

C.-T.

, & Huang

D.-S.

(2017). Asymmetric globalization and specialization. International Review of Economics & Finance , 52, 402–408. https://doi.org/10.1016/j.iref.2017.03.011

32.

T.H.

, Kim

, & Lee

(2016). Institutional quality, trade openness, and financial sector development in Asia: An empirical investigation. Emerging Markets Finance and Trade , 52(5), 1047–1059.

33.

X.L.

, Li

, & Si

D.K.

(2019). Asymmetric determinants of corporate bond credit spreads in China: Evidence from a nonlinear ARDL model. North American Journal of Economics and Finance , 7, 101109. https://doi.org/10.1016/j.najef.2019.101109

34.

Majeed

M.T.

, & Farooq

(2021). An empirical analysis of globalization–poverty nexus: Evidence from Pakistan. Review of Economics and Development Studies , 7(4), 533–542.

35.

Magombeyi

M.T.

, & Odhiambo

N.M.

(2017). Causal relationship between FDI and poverty reduction in South Africa. Cogent Economics & Finance , 5(1), 1–12. https://doi.org/10.1080/23322039.2017.1357901

36.

Marques

A.C.

, Fuinhas

J.A.

, & Tomás

(2019). Energy efficiency and sustainable growth in industrial sectors in European Union countries: A nonlinear ARDL approach. Journal of Cleaner Production , 239(42), 118045.

37.

Modebadze

(2012). Is globalization a source of prosperity and progress or the main cause of poverty? What has to be done to reduce poverty? Journal of Social Sciences , 1(1), 13–17.

38.

Musakwa

M.T.

, & Odhiambo

N.M.

(2019). The impact of remittance inflows on poverty in Botswana: An ARDL approach. Journal of Economic Structures , 8(1), 1–13. https://doi.org/10.1186/s40008-019-0175-x

39.

Nissanke

(2009). Linking economic growth to poverty reduction under globalization: A case for harnessing globalization for the poor in sub-Saharan Africa. AERC project (pp. 1–34).

40.

Nissanke

, & Thorbecke

(2008). Introduction: Globalization-poverty channels and case studies from sub-Saharan Africa . African Development Bank, Blackwell Publishing Ltd.

41.

Ogbuaku

N.H.

, Adebisi

, & Feridun

(2006). Does globalization lead to increased poverty in Africa? Empirical evidence from Nigeria. Economia Internazional Cmera di commercio di Genova , 59(1), 63 –82.

42.

Ojeyinka

T.A.

, & Osinubi

T.T.

(2022). The moderating role of financial development in the globalization-sustainable development nexus in some selected African Countries. Economic Change and Restructuring , 55, 2051–2080.

43.

Okungbowa

F.O.E.

, & Eburajolo

O.C.

(2014). Globalization and poverty rate in Nigeria: An empirical analysis. International Journal of Humanities and Social Science , 4(11), 126–135.

44.

Olaniyi

C.O.

(2022). On the transmission mechanisms in the finance–growth nexus in Southern African countries: Does institution matter? Economic Change and Restructuring , 55, 153–191.

45.

Orija

O.A.

, & Folawewo

A.O.

(2021). Does human capital matter for FDI’s effect on poverty? An evidence from Nigeria. Nigerian Journal of Economic and Social Studies , 63(1), 105–128.

46.

Osinubi

T.T.

(2020). The role of income inequality in the globalization–poverty nexus: Empirical evidence from MINT countries. Ilorin Journal of Economic Policy , 7(2), 67–89.

47.

Oyewale

I.O.

, & Amusat

W.A.

(2013). Impact of globalization on poverty reduction in Nigeria interdisciplinary. Journal of Contemporary Research in Business , 4(11), 475–484.

48.

Pesaran

(2003). A simple panel unit root test in the presence of cross-section dependence (Working Papers in Economics 0346). Faculty of Economics (DAE), University of Cambridge.

49.

Pesaran

M.H.

(2004). General diagnostic tests for cross section dependence in panels (Discussion Paper Series, DP No. 1240). IZA.

50.

Pesaran

M. H.

, Shin

, & Smith

R. J.

(2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics , 16(3), 289–326. https://doi.org/10.1002/jae.616

51.

Pesaran

M.H.

, Shin

, & Smith

R.P.

(1999). Pooled mean group estimation of dynamic heterogeneous panels. Journal of the American Statistical Association , 94(446), 621–634.

52.

Peres

, Ameer

, & Xu

(2018). The impact of institutional quality on foreign direct investment inflows: Evidence for developed and developing countries. Economic research [ Ekonomskaistraživanja ], 31(1), 626–644.

53.

Round

J.I.

(2007). Globalization, growth, inequality and poverty in Africa (Research Paper No. 2007/55) . UNU-WIDER.

54.

Salahuddin

, Vink

, Ralph

, & Gow

(2019). Globalization, poverty, and corruption: Retarding progress in South Africa. Development Southern Africa , 37(4), 617–643. https://doi.org/10.1080/0376835X.2019.1678460

55.

Salimono

(1999). Globalization and challenges . A paper presented at the international summit on globalization as a problem of development in Harana Cuba, North America.

56.

Sindzingre

(2005). Explaining threshold effects of globalization on poverty (Research Paper, 53) (pp. 1–27). WIDER.

57.

Shin

, Yu

, & Greenwood-Nimmo

(2014). Modelling asymmetric cointegration and dynamic multipliers in a nonlinear ARDL framework. In Sickles

, & Horrace

(Eds.), Festschrift in Honor of Peter Schmidt . Springer. https://doi.org/10.1007/978-1-4899-8008-3_9

58.

Singh

, & Huang

(2011). Financial deepening, property rights, and poverty: Evidence from sub-Saharan Africa (Working Paper No. WP/11/196). International Monetary Fund.

59.

Stark

(2004). Rethinking the brain drain. World Development , 32(1), 15–22.

60.

Stoyanova

, & Tonkin

R.P.

(2018). An expenditure-based approach to poverty in the UK [Paper presentation]. 35th IARIW General Conference, Copenhagen.

61.

Tsikata

Y.M.

(2001). Globalisation, poverty and inequality in sub-Saharan Africa: A political economy appraisal. A research programme on empowering people to meet the challenges of globalisation prepared for OECD Development Centre (Working Paper No. 183) (pp. 3–40). OECD.

62.

Ucal

M.S.

(2014). Panel data analysis of foreign direct investment and poverty from the perspective of developing countries. Procedia—Social and Behavioral Sciences , 109, 1101–1105.

63.

Uzonwanne

M.C.

(2018). Economic globalization and poverty reduction: A Nigerian perspective. Economics and Political Economy , 5(3), 410–423.

64.

Yıldırım

, Gedikli

, Erdoğan

, & Yıldırım

D.Ç

. (2020). Natural resources rents–financial development nexus: Evidence from sixteen developing countries. Resources Policy , 68, 101705.