Summary: The main objective of this study was to study the causal relationship between international remittance flows and poverty in South Africa, using time series data from 1980 to 2017. The study was motivated by two sources: the growing role of remittances in poverty reduction and human development, and the rapid growth in remittance inflows. Since 1998, remittance inflows to South Africa have shown a varying degree of upward trend. In 1998, for example, remittance inflows in the country increased by 18.5 per cent, and in the subsequent periods 1999-2017, remittance inflows increased steadily, with an average increase of 25.3 per cent. Our results using the multivariate Granger causality model show that when infant mortality is used as a proxy variable for poverty, there is a clear one-way causal flow from poverty to remittance in the short term. However, when household consumption expenditure was used as a proxy variable, no pervasive causal relationship was found in both the short and long term. This article discusses the implications of the policy. [Translated by Li Guanghui]
Author:Mersey M. Mercy T. Musakwa, Department of Economics, University of South Africa. P.O Box 392, UNISA 0003, Pretoria South Africa. Email: [email protected]
Nicholas M. Nicholas M. Odhiambo, Department of Economics, University of South Africa. P.O Box 392, UNISA 0003, Pretoria South Africa. Email: [email protected]; [email protected]
Source: Journal of Social Science (Chinese Edition), No. 4, 2021, P75-P86
Editor-in-charge: Liang Guangyan, Zhang Nancy
Introduction
The South African Government joined hands with other countries in the fight against poverty and signed the Millennium Development Goals (MDGs) (United Nations 2000) in 2000. This is a step towards South Africa's commitment to poverty eradication. While South Africa has achieved part of the Millennium Development Goals, the country is still battling poverty and inequality (Republic of South Africa 2015, p. 19). This has also led South Africa to participate in the Sustainable Development Goals (SDGs), which are the follow-up policies to the Millennium Development Goals (Millennium Development Goals) completed in 2015 (United Nations 2018). The primary objective of the two plans, spearheaded by the United Nations, is inclusive economic development. This policy is closely linked to the ability of countries to use resources at the national and international levels, including through FDI and recent remittance inflows. Remittances are becoming increasingly important as a source of financing for economic development, especially poverty alleviation, as reflected in Item 10.7 of the Sustainable Development Goals (United Nations 2017). While the United Nations has noted the important role that remittances can play in poverty reduction, empirical evidence on the relationship between remittances and poverty remains limited in sub-Saharan African countries, particularly in South Africa.
According to Ratha et al. (Ratha et al. Remittance inflows from low- and middle-income countries have increased significantly recently, reaching the $528 million mark in 2018. This is a 10.8% increase in remittance inflows from 2017 (Ratha et al. 2018）。 As a result, remittances have become an important source of external financing for low- and middle-income countries to compete with other external income streams, such as foreign direct investment. According to Ratha et al. (Ratha et al. 2018) study, if China is excluded from low- and middle-income countries, remittance inflows have grown to more than three times official development assistance (ODA) and outweigh foreign direct investment. More importantly, remittances have had a direct positive impact at the household level and indirect benefits at the national level.
There is already a wealth of literature on the impact of remittances on economic growth (see Goschin 2014; Atanda and Charles 2014; Meyer and Shera 2017; Makun 2018) is available. However, there is little literature on the impact of remittances on poverty (see Adam Jr. and Page 2005; Gupta et al. 2009; Adam Jr. and Cuecuecha 2013; Azam et al. 2016; Vacaflores 2018; Tsaurai 2018; Wangle and Devkota 2018）。 These studies did not reach definitive conclusions, and the results were different depending on the poverty proxy indicator used. The research literature on the impact of remittances on poverty reduction is very rich, but the causal relationship between remittances and poverty remains an area of obscurity, although this is important for government policy priorities (see, e.g., Abdulnasser and Gazi-Salah 2014; Gaaliche and Gaaliche 2014; Muhammad et al. 2016）。 Given the growing pressure on the South African Government to reduce poverty and, on the other hand, the increase inflow of remittances, research on the causal relationship between poverty and remittances will give policymakers insight into the macroeconomic variables of the target to achieve the desired poverty alleviation outcomes.
In light of the recent surge in remittance inflows, a new study exploring the causal link between remittances and poverty in South Africa will help policymakers make informed policy choices to make good use of remittances in the fight against poverty. Therefore, the main objective of this study was to investigate the causal relationship between poverty and remittances in South Africa.
This study differs from previous studies in that it uses two proxy indicators of poverty, namely household consumption expenditure (an income indicator of poverty) and infant mortality (a non-income indicator of poverty). First, the choice of these two proxy indicators has improved the robustness of the findings, especially given that the debate on the most comprehensive measure of poverty is still raging. Second, time series data for other poverty measures, such as the number of poor and the Human Development Index (HDI), are not available, which is why poverty proxy indicators were chosen. Ravallion (2001), Rehman and Shahbaz (2014) also used household consumption expenditure as a proxy indicator of poverty, while Abosedra et al. (Abosedra et al. 2016) and Odhiambo (Odhiambo 2011) used infant mortality as an agent indicator of poverty.
To the best of our knowledge, this study may be the first of its kind to study the dynamic causality between remittances and poverty reduction in South Africa using the ARDL (Autoregressive Division Lag) boundary test of Granger causality based on ECM (Error Correction Model).
The rest of the study is structured as follows: a literature review is discussed in the section on Remittances and Poverty Trends in South Africa, an estimation methodology is summarized in the Relevant Literature Review, the results are presented and discussed in The Estimation Methodology and Empirical Results, and the study is summarized in Conclusions and Policy Recommendations.
Remittances and poverty trends in South Africa
Remittances, as a source of economic development, have only recently received attention, in line with the thrust of efforts to achieve the Sustainable Development Goals. In recent years, more and more South Africans have been looking for a better future in other countries, although this number is still low compared to other countries such as India and Brazil (United Nations Conference on Trade and Development [UNCTAD] 2018). The top destinations for these South African migrants are Australia, the United Kingdom, the United States and New Zealand (Businesstech 2018). South Africa receives a significant number of migrants, mostly from African countries. As a result, government policies are more inclined to regulate remittance outflows.
From 1980 onwards, remittance inflows to South Africa gradually picked up, although the volume was still small (UNCTAD 2018). In 1980, remittances registered in South Africa accounted for 3% of GDP, then inflows increased slightly to 3.5% of GDP in 1984 (UNCTAD 2018). From 1989 onwards, there was a marked increase with registered inflows of 6.5% (UNCTAD 2018). Although the trend of inflows of 6.7% of GDP continued until 1992, South Africa experienced a slippery slope between 1993 and 1996 (UNCTAD 2018). In 1998, there was a 18.5% surge, and between 1999 and 2017, it maintained steady growth, averaging 25.3% (UNCTAD 2018). Remittance inflows were also the highest in the same period, at 29.1 per cent in 2009 (UNCTAD 2018).
South Africa has made concerted efforts to introduce pro-poor policies that allow both the Government and the private sector to play an important role. Poverty alleviation initiatives are embedded in the country's national economic policy, the National Development Plan 2030. In addition, the government's poverty alleviation initiatives have been deliberately made simple during the government's planning and budgeting process (South Africa Government 2008). The country's poverty reduction strategy is centred on eight pillars, namely: investment in human capital, income security, creation of economic opportunities, improved health care, access to assets, environmental sustainability, good governance, social inclusion and social capital initiatives, and basic services and other non-fiscal transfers such as subsidized housing, sanitation facilities and garbage removal (South Africa Government 2008). Within the overall framework of economic empowerment, the Ministry of Trade and Industry (DTI) has launched a number of programmes aimed at supporting formerly vulnerable groups in accessing the means of production. These programs include: Broad Black Economic Empowerment (B-BBEE); Industrial Innovation Support Program (SPII); Black Industry Program; Agroprocessing Support Program; and Seda Technology Program (DTI 2020; SME South Africa 2019）。 The Government's pro-poor policies take into account the most vulnerable groups in society — children, the elderly, the unemployed, the disabled, women and people living in deprived areas (South African Government 2008). In addition to focusing on long-term economic empowerment, governments are using social safety nets to meet the immediate needs of the poor, based on needs assessments. Some of the grants paid by the South African Government through the South African Social Security Service (SASSA) include: Child Support Grant, Senior Citizen Benefit, Social Assistance Grant, Veterans Grant, Nursing Dependency Grant, Disability Grant and Assistance Grant (SASSA 2020).
In response to these policy initiatives, South Africa has undergone a gradual process of poverty reduction, although poverty figures remain high and fluctuate over time (World Bank 2019). In 1993, the poverty gap was 29.3% and 9.5%, respectively, and then in 1996 poverty surged to 33.8%, with a poverty gap of 12.9% (World Bank 2019). From 2000 to 2010, both the number of poor and the gap between poverty and poverty continued to decline (World Bank 2019). In 2014, the number of people living in poverty in the country was 18.9%, an increase of 2.4% over 2010 (World Bank 2019). The poverty gap also surged from 4.9 percent in 2010 to 6.2 percent in 2014 (World Bank 2019). The Human Development Index (HDI) also increased by 0.081, from 0.618 in 1990 to 0.699 in 2017 (UNDP 2018). This is a remarkable achievement compared to the Human Development Index of 0.39 and 0.537 for sub-Saharan Africa in 1990 and 2017 (UNDP 2018).
Literature review
The United Nations identifies remittances as an important source of development finance for inclusive development, the Sustainable Development Goals (SDGs). This led to the inclusion of remittances in Section 10 of the Sustainable Development Goals list (United Nations 2018), which emphasizes the need to develop policies that support remittances between countries. Remittance is defined as a transaction initiated by an immigrant to a family member in their home country (United Nations 2018). There are different arguments as to why immigrants are willing to repatriate some of their income. According to Lucas and Stark (1985), migrant remittances have the following reasons: altruism, coinsurance and savings. At the heart of the altruistic motive is sympathy for the laggards and the need for economic assistance; coinsurance is the need to invest domestically so that if anything happens to them in a foreign country, they can go home; and the savings motivation is based on the drive that migrants use remittances as a means of savings in case of future investment or when foreign income flows slow or stop. Remittances can be in cash or in kind (Hagen-Zanker and Himmelstine 2016). Remittance to consumption (see Adam Jr. and Page 2005; Bui et al. 2015) and investment – real estate and small businesses (see Ratha 2007) have a direct impact.
Although this study focuses on the causal relationship between remittances and poverty at the household level, the indirect effects of remittances are achieved through a multiplier effect, which is initiated by initial increases in household consumption and investment. Thus, the indirect impact of remittances on poverty is felt at the national level, as it brings economic output to higher levels. In addition, the counter-cyclical nature of remittances makes it a good shock absorber in difficult times such as natural disasters and wars (Kapur 2004). The benefits of remittances can be summarized as poverty alleviation, stimulating economic growth, saving, stimulating sector growth, and investment (De Vries 2011).
Research on the causal relationship between poverty and remittances, while growing, remains extremely scarce. Most studies examining the relationship between poverty and remittances have focused on the impact of remittances on poverty. Studies on the impact of remittances on poverty are also divided into two types, one is the study that found positive effects (Adam Jr. and Page 2005; Gupta et al. 2009; Tsaurai 2018; Musakwa and Odhiambo 2019), another is the discovery of studies that determine whether relationships are sensitive or not by the poverty proxy indicators used (Azam et al. 2016; Wangle and Devkota 2018）。 Overall, the findings of studies investigating the impact of remittances on poverty support poverty alleviation effects. One of the few studies that examine the causal relationship between poverty and remittances can be divided into: Studies that discover two-way causality (Abdulnasser and Gazi-Salah 2014; Gaaliche and Gaaliche 2014; Sanchez-Loor and Zambrano-Monserrate 2015; Muhammad et al. 2016; Musakwa and Odhiambo 2020); study of one-way causality (Sanchez Loor and Zambrano-Monserrate 2015; Musakwa and Odhiambo 2020); and studies without causality (Muhammad et al. 2016; Sanchez-Loor and Zambrano-Monserrate 2015）。 Empirical commentaries on the study of causality between poverty and remittances provide an understanding of the nature of causal flows in other countries and provide a starting point for another study focused on South Africa. The results of this study will guide South Africa's policies on poverty reduction.
Gaaliche and Gaaliche 2014 used data from 1980 to 2012 to study the causal relationship between remittances and poverty in 14 emerging and developing countries. The two-way causal relationship between poverty and remittances has been confirmed. Abdulnasser and Gazi-Salah 2014 also used data from 1976-2010 to study the causal relationship between remittances and poverty in Bangladesh. Their findings are consistent with M.G. galiche and M. gaaliche (Gaaliche and Gaaliche 2014), both confirming a bidirectional causal relationship.
Sanchez-Loor and Zambrano-Monserrate 2015 used data from 1980 to 2012 to study causal flows between remittances and poverty in Colombia, Ecuador and Mexico. A two-way causal link between remittances and poverty was found in Colombia, a one-way causal flow from the Human Development Index (HDI) to remittances was demonstrated in Mexico, while no causality was found in Ecuador. Similarly, Muhammad et al. (Muhammad et al. 2016) using data from 1990-2014, the causal relationship between remittances and poverty in 39 low- and middle-income, upper-middle-income and high-income countries was studied. Using the Engle-Granger two-step test, they found that there is a one-way causal flow between foreign remittances and poverty in low- and middle-income countries. However, no causal link was found between poverty and foreign remittances in high-income countries.
Musakwa and Odhiambo 2020 used time series data from 1980 to 2017 to study the causal relationship between remittances and poverty in Botswana. The study used two poverty proxy indicators, household consumption expenditure and infant mortality. Using the autoregressive distributed lag (ARDL) method and the causality test based on the error correction model (ECM), a one-way causal flow from poverty to remittances was found in the long and short term. In the same study, when household consumption expenditure was used as a proxy indicator of poverty, a two-way causal flow between remittances and poverty was demonstrated.
Muhammad et al. (Muhammad et al. The 2016, Sanchez-Loor and Zambrano-Monserrate 2015 study confirmed inconsistent results and the sensitivity of these results to the countries studied. From these studies, it can be concluded that it is inappropriate to generalize the results of one country to another. Given the uncertain outcome, only empirical studies of the relationship between poverty and remittances in South Africa can provide additional insights.
Estimation methods and empirical results
This study is based on the ARDL boundary test and the ECM-based causality test. Ardl testing was chosen because of its many advantages. First, the ARDL method has no restrictions on the order in which the variables contained in the model can be integrated. This method allows the combination of variables of order 0, 1, or partial integration (Pesaran et al. 2001）。 However, the method fails when corroborating variables of order 2 or higher. Second, the method is robust even in small sample sizes (Solarin and Shahbaz 2013). Third, other traditional cointegration methods use systems of equations, while the ARDL method uses a simplified, easy-to-handle single equation (Pesaran and Shin 1999).
Corroborating confirms that there is a long-term relationship between the variables in the model. The existence of cointegration indicates only the existence of a long-term relationship and the existence of a causal relationship in at least one direction (Narayan and Smyth 2008). In models that confirm cointegration, error correction models are used to estimate short- and long-term causal relationships. Those models that do not confirm the cointegration relationship estimate only short-term causal relationships. The causal relationship between poverty and remittances was studied using an ECM-based approach within a multivariate framework. In addition to remittances and poverty, two other variables included in the model are real gross domestic product (GDPC) per capita and education (EDU). Both factors are important for both remittance inflows and poverty. Given that the weakness of the bivariate framework is that the model may be affected by the omission of variable bias, the inclusion of these two variables minimizes the impact (Odhiambo 2008). A framework with more than two variables can increase the amplitude of the result (Odhiambo 2009a).
Some of the poverty proxy indicators used in this study include household consumption expenditure, GDP per capita, infant mortality and life expectancy, and other poverty proxy indicators. In view of the limited time series data, this study used household consumption expenditure (Pov1) and infant mortality rate (Pov2) as proxy indicators for poverty, reflecting the income and non-income levels of poverty, respectively. The two models used are: Model 1 is represented as Pov1| REM, GDPC, EDU; Model 2 is expressed as Pov2| REM, GDPC, EDU。
The definition of the variable
The study used two poverty proxy indicators: household consumption expenditure and infant mortality. Household consumption expenditure is measured as a percentage of gross domestic product (GDP). This poverty proxy indicator has been used as a measure of poverty in some studies (see Rehman and Shahbaz 2014; Magombeyi and Odhiambo 2017; Kaidi et al. 2018; Musakwa and Odhiambo 2020）。 This study selected household consumption expenditure to reflect the income dimension of poverty. The advantage of using expenditure measures rather than pure income measures is that expenditure measures are less sensitive to changes in income, especially when these changes are only temporary (Warr 2006). The one-way causal flow from poverty (household consumption expenditure) to remittances means that high levels of household consumption expenditure lead to more remittance inflows. If there is a one-way causal flow from remittances to poverty, the opposite is true. Thus, remittances inflows increase household consumption expenditure and, consequently, reduce poverty. Infant mortality is measured by the number of deaths per 1,000 live births (World Bank 2019). In some studies, this proxy indicator has been used to measure poverty (e.g., see Abosedra et al. 2016; Van Multzahn and Durrheim 2008）。 According to the classification of health measures by Carvalho and Horward (1996), infant mortality is one of the social indicators used to measure health status. High infant mortality indicates poor health in countries, while countries with low infant mortality imply access to health services. The one-way causal flow from infant mortality to remittances means that poverty leads to remittances flowing in, while the one-way causal flow from remittances to infant mortality means that remittances contribute to poverty reduction.
Remittances are measured as a percentage of GDP. The advantage of this is that the size of the country is taken into account. Real GDP per capita measures a country's economic development, taking into account the size of the population. Assuming an equitable income distribution, this is a better measure of the standard of living that a country's population generally enjoys. Assuming a fair distribution of income, the higher the per capita income, the better the living conditions of the people of the country. The measure of education is the gross enrolment rate at the primary level. Given the limited data on enrolment in secondary education since 1980, the study used gross primary school enrolment to measure the level of human capital. A number of human capital substitution indicators have been used in the literature, including primary school gross enrolment, secondary school gross enrolment and gross enrolment (see, e.g., Borreinzstein et al. 1998; Jalilian and Weiss 2002; Gohou and Soumare 2012）。 High enrolment rates are expected to produce a high-quality workforce, as well as relatively high incomes. This income translates into a higher standard of living, allowing the poor to improve their economic situation.
According to Narayan and Smyth 2008, the specific description of the ARDL boundary of Model 1 and Model 2 is shown in Equations 1-4:
General cointegration models (Povm, REM, GDPC, EDU)
Where, when m=1, Povm's position in Model 1 is Pov1 (household consumption expenditure); when m=2, Povm's position in Model 2 is Pov2 (infant mortality); they enter the equation one at a time, rem is remittances as a percentage of GDP; EDU is education, GDPC is real GDP per capita, and α0, μ0, π0, β0 are constants, and the range is:
And β1-β8 is the regression coefficient; μ1-μ4 is the error term.
After the cointegration test, the next step is to examine the direction of causation. This is done through the ECM-based Granger-causality model. Models 1 and 2 are specified in the ECM-based causality model, as shown in Equations 5-8. ECM-based causality testing allows for the analysis of short- and long-term causal relationships. Short-term causation is tested using the F-statistic from the variable loss test, while long-term causation is obtained from the t-statistic in the lag error correction term.
Granger's causality model based on the general ECM is specified, as shown in Equations 5-8.
thereinto
is a constant;
is the regression coefficient; all other variables, as described in Equations 1-4.
Data sources
This study used time series data from 1980 to 2017 to study the causal relationship between remittances and poverty. Time series data on education, infant mortality, household consumption expenditure and GDP per capita were extracted from World Bank development indicators. Remittance data are taken from the United Nations Conference on Trade and Development (UNCTAD) database. The data analysis was made using Microfit 5.0 software.
Unit root test
The study conducted root-of-unit tests for Pov1, Pov2, remittances (REM), real gross domestic product per capita (GDPC), and education (EDU). Although no pre-test is required when using the autoregressive distributed lag (ARDL) boundary test method, the purpose of the test is to determine whether the variable is a order integral [I(1)] or a order integral [I(0)]. Ardl (Pesaran et al. can only be used if the variables are [I (0)] or [I (1)] or partially integrated. 2001）。 The results of the Dickey-Fuller Universal Least Squares Test (DF-GLS) and the Philippe-Perone Test (PP) are shown in Table 1.
The results in Table 1 confirm that all variables are stationary at first-order differentials. The next step is to test the cointegration. The variables incorporating the cointegration function are Pov1, Pov2, REM, GDPC, and EDU. In Model 1, Pov1 is used as a poverty proxy indicator, while in Model 2, Pov2 is used as a poverty proxy indicator. The remaining variables in both models remain unchanged. The cointegration result is shown in Table 2.
The results of Table 2 confirm the cointegration of some of the functions in Model 1 and Model 2. The calculated F-statistic is related to Pesaran et al. (Pesaran et al. 2001) provides a comparison of the cut-off values. If the computed F-statistic is greater than the upper critical value, the cointegration relationship is confirmed, while if the F-statistic is below the lower critical value, the cointegration relationship is not confirmed. When the F-statistic is between Pesaran et al. (Pesaran et al. 2001) when provided between the lower and upper bounds, the test results are indeterminate. In Model 1, the existence of cointegration is confirmed only in the Pov1 function, while in Model 2, the cointegration is confirmed in both the Pov2 and EDU functions. Cointegration is confirmed in the following functions: Model 1, F(Pov1| REM, GDPC, EDU); Model 2, F(Pov2| REM, GDPC, EDU) and F (EDU| REM，GDPC，Pov2）。 The existence of cointegration indicates a causal relationship in at least one direction (see Narayan and Smyth 2008). In order to determine the direction of causation, this study used an ECM-based causality test. The results of the ECM-based causality test are shown in Table 3.
The results in Table 3 confirm the one-way causal flow from Pov2 (infant mortality) to remittances in the short term. This is confirmed by the F-statistic of short-term causal flow, which has a significance of 5%. These results show that in South Africa, the Granger causality test results in infant mortality leading to remittance inflows. In South Africa, a possible explanation for this relationship is that migrants must help their families back home, especially when they are poor. According to the altruistic and coinsurance incentives of remittances, the more difficult the family, the more likely it is that migrants will send remittances back home to alleviate economic hardship (Lucas and Stark 1985, p.94; Depoo 2014, p.203）。 These findings suggest that poverty levels play a key role in harnessing remittances. When household consumption expenditure was used as a proxy indicator, no causal relationship was confirmed in the short or long term. Thus, the causal relationship between poverty and remittances is sensitive to the proxy indicators examined. With other studies such as Muhammad et al. (Muhammad et al. The results of this study are advantageous over the 2016, Sanchez-Loor and Zambrano-Monserrate 2015 study.
Other empirical findings in group A of table 3 show that the following situations exist in South Africa. (1) A one-way causal flow from education to real gross domestic product (GDPC) in the short term. This may be attributed to the fact that educated individuals are highly productive and innovative, which has had a positive impact on GDPC. (2) There is a two-way causal relationship between education and household consumption expenditure in the short term, and there is a one-way causal flow between education and household consumption expenditure in the long term. These results were confirmed by the F and t statistics of the lagging ECM, respectively. The more educated an individual is, the greater the chance of getting a better-paying job, which increases income and has a positive impact on consumption. (3) There is a two-way causal relationship between GDPC and remittances in the short term. (4) There is no confirmed causal relationship between household consumption expenditure (HHC) and GDPC in the short term and in the long term. (5) There is a one-way causal flow between remittances and education in the short term. These findings are supported by the theoretical claim that remittances lead to consumption and increased investment in education and assets (see Adam Jr and Page 2005; Ratha 2007; Bui et al. 2015）。
The empirical results of Group B in Table 3 show that there is a two-way causal relationship between remittances and education in South Africa: (1) there is no causal relationship between remittances and education in the short and long term; (2) GDPC triggers remittances in the short term, which is confirmed in the F statistic of 10% significance; (3) in the short term, there is a two-way causal relationship between GDPC and Pov2 (infant mortality), and there is a one-way causal flow from GDPC to infant mortality in the long term; (4) in the short term, There is a two-way causal relationship between GDPC and education, and there is a one-way causal flow between GDPC and education in the long run; (5) there is a one-way causal relationship between Pov2 (infant mortality) and education in the short term and in the long term. These findings on the causal relationship between economic growth and poverty are consistent with those of Odhiambo (2009b) and Pradhan 2010, who found that economic growth is a cause-and-effect factor in Granger's poverty reduction.
Table 4 gives a summary of the results of Granger's causal relationship.
Conclusions and policy recommendations
Using data from 1980-2017, this study investigated the causal relationship between remittance inflows and poverty in South Africa. The motivation for the study is that, on the one hand, the growing role of remittances in poverty reduction and human development has increased dramatically. This study used the ECM-based Granger causality model to examine this connection. To minimize the variable omission bias found in some previous studies, real GDP per capita and education variables were used as control variables, resulting in a multivariate Granger causality model. To improve the robustness of results, we used two poverty proxy indicators, household consumption expenditure and infant mortality. The results of the study show that when infant mortality is used as a proxy indicator of poverty, the results show that poverty affects granger-induced remittance inflows in the short term. However, when household consumption expenditure was used as a proxy indicator, no causal relationship was found between poverty and remittances in South Africa. This applies, whether it is a short-term or long-term assessment. Therefore, the study concludes that in South Africa, the causal relationship between remittances and poverty is sensitive to proxy indicators used to measure poverty levels. Based on the results of this study, we recommend that South Africa should continue to promote policies to reduce the cost of migration as a way to encourage migration. Such a policy stance would therefore encourage migration and increase remittance inflows. In addition, South Africa should work with major destination countries to rationalize policies to reduce the cost and time it takes for migrants to repatriate money to South Africa. This helps to increase the volume of remittances after deducting the cost of remittances, enables recipients to receive remittances in a timely manner to meet their urgent needs and encourages senders to use formal channels, which in addition to making it easier to collect statistics for reporting purposes. However, the neutral causal relationship between household consumption expenditure and remittances suggests that remittances are more sensitive to healthy poverty than income poverty. Given the nature and complexity of poverty in developing countries, this finding is not surprising. It is therefore recommended that South Africa consider a multi-pronged approach to poverty reduction that adequately addresses the multifaceted nature of poverty.