Introduction

Remittances sent from abroad to developing countries have grown significantly not only in absolute terms but also in importance regarding their share in gross domestic product. They are an importance source of income for poor households who usually rely on them for current consumption, education and investment. At the macroeconomic level, remittances are the second largest source of external financing after foreign direct investment (Aggarwal et al. 2011; World Bank 2014; Bhattacharya et al. 2018). Statistics from the World Bank point out that over the past 30 years remittances to Sub-Saharan Africa (SSA) increased from 0.8% of GDP in 1990 to 1.3% in 2000 and reached 2.8% in 2019, and then declined to 2.5% in 2020 following the coronavirus (COVID-19) disease. The actual figure is much higher, because a large portion of remittances to SSA is not recorded. According to the World Bank (2006), the total amount remitted could be as much as 50% higher than formal remittance inflows. Freund and Spatafora (2005) also reported that the amount of informal remittances to SSA accounts for about 45–65% of official flows. In many cases, remittances to SSA are sent through informal channels to avoid high fees for money transfer.

Not surprisingly, the increasing trend in remittance inflows to developing countries has motivated researchers to investigate their potential contributions to economic development. A growing pool of studies has analyzed the impacts of remittances on various dimensions of development, including poverty, inequality, consumption, human capital, investment, financial development, trade balance and economic growth (e.g., Adams and Page 2003; Edwards and Ureta 2003; Gupta et al. 2009; Salas 2014; Coulibaly 2015; Akobeng 2016; Keho 2017; Azizi 2018; Eggoh et al. 2019). Remittances can promote economic growth through increasing consumption expenditure and investment in the home countries particularly in countries that are characterized by underdeveloped financial sector and credit constraints. Much of existing empirical studies on remittances and financial development concentrate on their growth effects. Consequently, remittances and financial development have been identified as major drivers of economic growth (e.g., Levine et al. 2000; Hassan et al. 2011; Afawubo and Fromentin 2013; Keho 2020).

Since the bulk of remittance inflows to developing countries are sent through informal channels, financial sector development is viewed as an important link between remittances and investment. The financial sector can play a key role in channeling remittances into savings and investments. A number of studies have documented empirical evidence on the link between remittances and financial development in Africa. It has been shown that remittances thrive in countries with well-developed financial system. If a significant portion of remittances is transferred and saved through financial sectors and channeled into investment, then financial sector may contribute to enhance the growth effect of remittances through domestic capital formation. Even if remittances are spent on consumption, aggregate demand would increase, thereby stimulating investment in productive sector through the multiplier effect. Hence, it is important to understand the link between remittances and investment while considering the level of financial development in the recipient country. Recently, the investigation of the effects of remittances and financial development on investment has been a subject of research. However, little attention has been devoted to scrutinize the joint effect of remittances and financial development on investment, particularly in Sub-Saharan Africa.

This study fills the gap in the empirical literature by examining the effect of remittances on domestic investment using financial development as a transmission channel, with a focus on West African countries. The purpose of the study is to scrutinize whether financial sector development modulates the effect of remittances on domestic investment. More precisely, we test the complementarity or substitutability relationship between financial development and remittances in driving investment. The literature suggests that complementarity and substitution may exist between remittances and financial development in their relationship with domestic investment. The complementarity hypothesis demonstrates that remittances and financial development foster one another in their impact on investment. When remittances are sent through the financial system, they increase savings and thereby the ability of the financial sector to provide more credit to the private sector for investment. Conversely, the substitutability hypothesis suggests that remittances act as a substitute of domestic credit in less-financially developed countries. Better understanding the effect of remittances on investment under different levels of financial development would help in formulating appropriate policy to enhance the overall impact of remittances and financial sector development on economic growth in developing countries.

This study contributes to the existing literature in four ways. First, it evaluates the direct effects of remittances and financial development on domestic investment. Second, the study tests the existence of complementary or substitution effect between remittances and financial development in their relationship with domestic investment. Third, in terms of methodology, the present study deploys the Pooled Mean Group (PMG) and Mean Group (MG) estimators, as opposed to traditional panel estimation methods, to address the issues of stationarity, endogeneity and heterogeneity. Finally, we focus on West African countries, a sub-region rarely covered by recent research, but that receives the majority of the remittance inflows to Sub-Saharan Africa. By looking specifically at Western Africa, the research achieves a richer examination of the role of remittances in the sub-region than that covering SSA. To the best of our knowledge, this study is the first of its kind leading the way in testing the mediating role of financial development in the remittances and investment nexus in West African countries.

The rest of the study is organized as follows. The second section deals with the literature review. The third section describes the trend in remittance inflows to the countries under study. "Model, methodology and data" section outlines the estimation strategy and describes the data used in the empirical analyze. "Results and discussion" section discusses the empirical results of the study, while "Conclusion and recommendations" section concludes the study and provides some policy recommendations.

Literature review

The role of remittances in economic development has stimulated a growing literature. On the theoretical ground, the effect of remittances on economic growth depends on whether they fund consumption or investment expenditures whether they encourage work or leisure activities (Bettin and Zazzaro 2012). A pessimistic view argues that remittances to develop countries could not stimulate economic growth, because they are predominantly used for basic consumption needs rather than productive investment. Remittances could also hamper economic growth when they reduce the motivation of recipients to work. On the other hand, an optimistic view argues that remittances could improve economic growth directly through savings and investment in physical and human (education and health) capital, and indirectly through consumption smoothing and deepening financial market (Ratha 2003; Combes and Ebeke 2011; Mondal and Khanam 2018). On the empirical ground, an overwhelming body of literature has quantified the growth effect of remittances in individual countries or group of countries. The findings have been mixed, ranging from negative effect to positive effect.

Direct impact of remittances on investment

In a survey of 538 estimates reported in 95 studies, Cazachevici et al. (2020) found 40% of the studies reporting a positive effect, 40% showing no effect, and 20% reporting a negative effect. In a study for 113 countries, Chami et al. (2005) established negative effects of remittances on economic growth. This is because remittances are associated with reduction of labor force participation and moral hazard problems. Kagochi et al. (2010) examined the relationship between remittances and economic growth in a cross-country panel data analysis of six Sub-Saharan Africa (SSA). They found that remittances have a positive impact on economic growth in countries with high GDP per capita but do not cause direct impact on economic growth in low GDP per capita countries. Baldé (2011) analyzed comparatively the impact of remittances and foreign aid on savings and investment in 37 and 34 Sub-Saharan African countries over the period 1980–2004. The results showed that both remittances and foreign aid increase significantly savings and investment in SSA. Furthermore, remittances are more effective than foreign aid. In 36 SSA countries during the period 1990–2008, Singh et al. (2011) reported a negative effect of remittances on economic growth. Ahamada and Coulibaly (2013) explored the link among remittances and economic growth in 20 SSA countries over the period 1980–2007. They disclosed that there is no causal link between the two variables. Remittances do not enhance economic growth in SSA countries, because they do not increase physical capital investment. Hossain and Hasanuzzaman (2013) applied the ARDL bounds testing approach to the remittances and investment nexus in Bangladesh. The findings show that remittances positively and significantly influence the level of investment. Using data from 1970 to 2010 for 89 developing countries, Ngoma and Ismail (2013) found that remittances promote human capital development. Salahuddin and Gow (2015) applied the Pooled Mean Group regression technique to examine the growth effect of migrant remittances over the period 1977–2012 for four of the largest recipient countries of foreign remittances namely, Bangladesh, India, Pakistan and the Philippines. They found a positive growth effect of remittances in the long-run in these countries. Yiheyis and Woldemariam (2016) looked at the remittances and domestic investment relationship in four SSA countries (i.e. Burkina Faso, Nigeria, Kenya, and Senegal) and reached differing impacts, with significant negative effects in two cases. Tung (2018) depicted the impact of remittances on domestic investment for 19 Asia–Pacific countries over the period 1980–2015. Results based on panel fixed effect ordinary least squares and two-stage least squares regressions disclosed that remittances have a negative impact on domestic investment. Conversely, GDP per capita growth, domestic credit and gross saving have a positive impact on domestic investment. Eggoh et al. (2019) used a panel of 49 developing countries from 2001 to 2013 and concluded to a positive effect of remittances on economic growth. Adekunle et al. (2020) applied the ARDL approach to evaluate the interaction of remittances and institutions with economic growth in Nigeria. Remittances were found to be growth-retarding in the long-run. However, improved institutional environment offset this adverse impact. Bird and Choi (2020) examined the impact of remittances, foreign direct investment and foreign aid on economic growth in 51 low-income and middle-income developing countries over the period 1976–2015. They found a negative relationship between remittances and economic growth whereas foreign direct investment has a positive effect. Ekanayake and Moslares (2020) analyzed the effects of remittances on economic growth in 21 Latin American countries for the period 1980–2018. They found that remittances contribute positively to economic growth. Omon and Ahuru (2020) investigated the impact of remittances on economic growth for the member countries of West Africa Monetary Zone (WAMZ) over the period 1990–2016. The results from Pooled Mean Group (PMG) estimator showed that remittances have a negative relationship with real per capita gross domestic product in the long-run. This may be due to the fact that remittances are not spent on investment projects. Haque et al. (2021) investigated the effects of remittances on saving and investment rates in five South Asian countries over the period 1985–2018. The results from OLS and 2SLS fixed effects methods showed that remittances have a positive impact on saving rate but have no significant effect on investment rate. The study concluded that remittances are mostly used for meeting consumption and non-productive purposes in south Asia. Odugbesan et al. (2021) examined the effect of financial development and remittances on economic growth in MINT nations (Mexico, Indonesia, Nigeria, and Turkey) over the period from 1980 to 2019. The results from panel linear ARDL and panel nonlinear ARDL suggest that both financial development and remittances promote economic growth. Nyasha and Odhiambo (2022) unlocked the impact of remittances on economic growth in South Africa over the period 1970–2019. Using the ARDL bounds testing approach, they reported that remittances have a negative impact on economic growth both in the long and short run.

Complementarity and substitutability between remittances and financial development

Even though there has been a great deal of research with regard to remittances, financial development and investment, most investigations have focused on their separate effects. Given mixed evidence from existing literature, another strand of the literature emphasizes the role financial sector development plays in the nexus between remittances and investment with the aim of testing the complementarity or substitutability hypothesis. The complementarity hypothesis claims that remittances and financial development foster one another. By increasing savings, remittances increase the ability of financial sector to provide credit to the private sector and this may impact positively on investment. On the other hand, the substitution mechanism works in less-financially developed countries, where remittances compensate for inefficient credit markets by providing an alternative finance source for investment (Giuliano and Ruiz-Arranz 2009; Fayissa and Nsiah 2010; Bettin and Zazzaro 2012). By sending remittances, migrants play the role of financial intermediaries, enabling recipients to overcome credit constraints when they intend to invest in human and physical capital. This reduces formal credit demand and impedes the development of financial sector. Consequently, remittances are likely to have more detrimental impact on domestic investment in countries with under-developed financial sectors.

A number of empirical studies have analyzed the role of financial development in the remittances and growth or investment nexus. Giuliano and Ruiz-Arranz (2009) conducted a study for 73 developing countries over the period 1975–2002. They reached the conclusion that credit to private sector and remittances are substitute in their relationship with economic growth. The work by Mundaca (2009) used a dataset of 39 Latin American and Caribbean countries over the period 1970–2002 and found that remittances are growth-enhancing. The study also found evidence of complementarity between remittances and financial depth in triggering economic growth. Fayissa and Nsiah (2010) investigated the growth impact of remittances for a panel of 36 African countries over the period from 1980 to 2004. They found that remittances enhance economic growth in countries where the financial system is sound. Nyamongo et al. (2012) confirmed the complementarity hypothesis between remittances and financial development for a panel of 36 Sub-Saharan African countries. In a study of 36 SSA countries over the period 1990–2008, Lartey (2013) found that remittances improve economic through investment and consumption smoothing. The results further showed a positive interaction effect between remittances and financial depth. Ojapinwa and Odekunle (2013) explored the impact of remittances on investment and their interaction with financial development in Nigeria. They found that remittances increase the stock of physical investment. Furthermore, financial development complements remittances in spurring capital formation. Chowdhury (2016) examined how financial development influences the growth effect of remittances in top 33 remittance receiving developing countries over the period 1979–2011. Applying a dynamic panel estimation method, remittances were found to significantly promote economic growth, while the effect of financial development is insignificant. Furthermore, the interaction effect of financial development and remittances is not significant. Therefore, financial development neither works as a substitute nor a complement for the remittance–growth nexus. Fowowe and Ibrahim (2016) investigated the effects of remittances on economic development in Lesotho during the period 1970–2014, focusing on the role of financial development. They used the fully modified OLS estimation technique. The results indicated that both remittances and financial development have positive effects on per capita GDP. However, remittances and financial development are substitutes in affecting per capita GDP. El Hamma (2018) scrutinized the effect of remittances, financial development and institutional quality on economic growth for 14 Middle East and North Africa (MENA) countries over the period 1982–2016. Using Two-Stage Least Squares instrumental variables method, the results reveal a complementary relationship between financial development and remittances in bringing about economic growth, suggesting that remittances improve growth in countries with a developed financial system. Kratou and Gazdar (2018) examined the effect of remittances and financial development on economic growth in a panel of 30 African countries over the period of 1988 to 2012. Using the system Generalized Method of Moments (GMM), results showed a complementarity between financial development and remittances in enhancing economic growth. Bandura et al. (2019) applied the GMM dynamic panel techniques to the case of 14 SADC member countries over the period 2006–2016. They found a positive impact of remittances on economic growth. Furthermore, a negative association between remittances and financial development was found, lending credence to the substitution hypothesis. Githaiga (2019) investigated the impact of foreign remittances and banking sector development on private investment in a sample of 15 Sub-Saharan African countries over the period 1986–2017. The findings indicated that both foreign remittances and banking sector development have a positive and significant effect on private investment in Sub-Saharan Africa. Moreover, the banking sector development has a negative moderating effect in the sense that an improvement in banking sector development diminishes the effect of remittances on private investment. Olaniyan (2019) guessed the joint impact of remittances and financial development on economic growth in Nigeria for the period 1977–2017. He reported a negative impact on remittances but a positive interactive effect of remittances and financial development, implying that financial development positively modulates the effect of remittances on economic growth. Olayungbo and Quadri (2019) scrutinized the nexus between remittances, financial development and economic growth in a panel of 20 Sub-Saharan African countries over the period 2000–2015. Using both Pooled Mean Group and Mean Group estimations, they found that remittances and financial development have positive effects on economic growth. The negative interactive effect suggests that financial development and remittances substitute for each other. Peprah et al. (2019) examined the case of Ghana over the period 1984–2015 through the ARDL approach. They found that remittances boost economic growth and that remittances and financial development interaction is growth-inducing. The results from quadratic model revealed the existence of a financial development threshold effect, indicating that an excessive expansion of the financial sector above 70% has a declining effect on economic growth. Sobiech (2019) reported evidence showing that remittances enhance economic growth, but the effect is significant only at low levels of financial development. Bangake and Eggoh (2020) tested for threshold effect of financial development in the remittances and economic growth nexus in a panel of 60 developing countries from 1985 to 2015. Using threshold regression model, they found that beyond a given threshold of financial development, remittances improve economic growth, while the effect is insignificant under the threshold. In other words, the growth-enhancing effect of remittances works only in recipient countries with relatively well-functioning financial sectors. Cao and Kang (2020) validated the substitute relationship between remittances and financial development in promoting economic growth for 29 economic transition countries during the period of 2000–2015. Dash (2020) investigated the impact of remittances on domestic investment for a panel of six South Asian countries over the period 1991–2017. Applying the GMM system method, they found that both remittances and financial development increase domestic investment rate. Furthermore, financial development has a complementary effect on the use of remittances for investment. Ngoma et al. (2021) also reported evidence confirming the complementary hypothesis in a panel study of Asian countries. Rehman and Hysa (2021) analyzed the experience of six Western Balkan countries during the period from 2000 to 2017. Using the system GMM method, they reported that financial development and remittances have positive impact on economic growth. However, the interaction of financial development and remittances showed a significant and negative effect, giving credence to the substitution effect. Ali Bare et al. (2022) examined the impact of remittances on human capital investment and the mediating role of financial development in a panel of 41 SSA countries over the period 1996–2016. The results revealed that remittances increase human capital investment, and the impact is more pronounced in countries where the financial sector is sound. Dash (2022) examined the relationship between remittances and domestic investment for 24 low-income countries over the period 2004–2018. Using second-generation panel estimation methods that account for endogeneity problem and cross-sectional dependency among countries, remittances were found to crowd in domestic investment, the crowding in effect being more pronounced in countries with more developed financial systems and higher human capital levels. Nyeadi et al. (2022) explored the experience of a panel of 41 African countries from 2004 to 2018. They reported that remittances have a negative impact on domestic investment whereas banking sector development positively impacts investment. Furthermore, banking sector development was found to improve the effect of remittances on domestic investment.

Motivation and hypotheses of the study

The empirical literature reviewed above clearly shows that the relationship between remittances, financial development and investment is still an infrequent topic. Existing studies have extensively focused on the individual effects of remittances and financial development on economic growth or investment. Little attention has been devoted to the interaction effect between remittances and financial development in affecting investment in SSA countries. In addition, the existing studies on the effect of remittances and financial development on investment reveal disparate outcomes, warranting for further research. The present study contributes to the literature by investigating the mediating role of financial development in the relationship between remittances and investment for a panel of 10 West African countries selected on the basis of data availability, namely, Benin, Burkina Faso, Cote d’Ivoire, Ghana, Mali, Niger, Nigeria, Senegal, Sierra Leone and Togo. Based on the literature reviewed above, the study tests the following hypotheses.

H1:

Financial development exerts a positive effect on domestic investment in West Africa.

H2:

Remittances have a positive effect on domestic investment in West Africa.

H3:

Financial development enhances the effect of remittances on domestic investment in West Africa.

The conceptual framework for the study is portrayed graphically in Fig. 1. It shows the linkages among remittances, financial development, and domestic investment.

Fig. 1
figure 1

Conceptual diagram

Full size image

The first hypothesis suggests that remittances have direct and indirect effects on investment. In perspective of his return to home, migrant sends money to home for investment in sectors, such as housing and health as well as small enterprise development. Furthermore, migrant helps his family members to start productive activities in order to prevent them against adverse economic chocks. When remittances are spent on consumption, they stimulate aggregate demand, and by the working of the multiplier, they contribute to trigger domestic investment and growth. The second hypothesis signifies that financial development channels savings into productive investment as advocated by the theory of financial development developed by Schumpeter (1911), Mckinnon (1973) and Shaw (1973). The rationale behind the third hypothesis is that by improving access to financial services and easing credit constraints, financial sector development attracts more official remittances and allows greater credit and investment. As reported by Freund and Spatafora (2005) and World Bank (2006), a large share of remittance inflows to SSA is sent through informal channels and is not recorded in official statistics. The development of the financial sector increases migrant transfers through official channels by lowering transaction costs and improving service availability (Aggarwal et al. 2011; Bettin et al. 2012). As the level of financial development expands, the amount of remittances will increase. Therefore, we expect financial development to strengthen the positive effect of remittances on investment that is, remittances are more conducive to investment in countries with higher level of financial development.

Trends and significance of remittances for West African countries

The trend of remittance inflows to the selected African countries is presented in Fig. 2. It shows that these countries witnessed an impressive increase in remittance inflows which increased from mere USD 0.38 billion in 1985 to USD 2.1 billion in 2000 and further to about USD 33.3 billion in 2019. Overall, the remittances inflows grew at an average rate of 16% per annum, mainly driven by Nigeria, Ghana and Senegal. It is worth noting that remittances increased during the period corresponding to the global financial crisis between 2007 and 2008. Remittance inflows rose from USD 20.6 billion in 2007 to around USD 22.2 billion in 2008, which represents a growth rate of 8%.

Fig. 2
figure 2

Source: World Development Indicators of the World Bank (2021). Data are in current USD billion

Trend of remittance inflows to selected West African countries from 1985 to 2019.

Full size image

Table 1 shows the trend when remittances are expressed as a percentage of gross domestic product. As this table shows, remittance inflows to West African countries increased from 2.2% of GDP in 1985 to 5.5% of GDP in 2019. This surge was observed in almost all countries. Remittance earnings constitute 10.70% of GDP for Senegal, 8.35% for Togo, 6.05% for Ghana and 5.98% for Mali. Domestic credit to the private sector and domestic investment have also increased from 1985 to 2019.

Table 1 Remittance inflows, domestic credit and investment (% GDP)
Full size table

Model, methodology and data

Empirical model

To empirically assess the impact of financial development and remittances on aggregate capital formation, we specify the empirical model as follows:

$${\text{I}}_{{{it}}} = \beta _{{0{\text{i}}}} + \beta _{{{\text{1i}}}} {\text{FD}}_{{{{it}}}} + \beta _{{{{2i}}}} {\text{REM}}_{{{{it}}}} + \beta _{{{\text{3i}}}} {\text{X}}_{{{{it}}}} + \mu _{{{{it}}}}$$
(1)

where Iit denotes total investment in country i in year t; FDit is financial development indicator; REMit represents remittances; Xit is a set of control variables, such as real GDP per capita (Yit), trade openness (OPit) and inflation rate (INFit). In addition, µit is an error term normally distributed. An important feature of our model is that it does not impose a common coefficient on each explanatory variable. We are interested in testing whether the effects of remittances (REM) and financial development (FD) are statistically significant.

Financial development is expected to have a positive impact on domestic investment by providing more credit for financing productive investment (i.e., β1 = δI/δFD > 0). The coefficient on REM is expected to be positive, since remittances provide external resources for funding investments (i.e., β2 = δI/δREM > 0). An increase in real income is likely to stimulate the aggregate demand, which in turn will enhance domestic investment. Therefore, the coefficient of real GDP per capita is expected to be positive. The effect of trade openness is expected to be positive as trade leads to higher technology and knowledge spillovers and enhances access to imported inputs. This positive effect could, however, be offset if increased access to imports discouraged domestic production. Finally, inflation as a proxy for macroeconomic stability for the economy, is expected to discourage investment.

Beyond the direct impact of remittances on investment described in Eq. (1), their effect cTo empirically assess an occur through financial development. We estimate the joint effect of remittances and financial development on investment through the following model:

$${\text{I}}_{{{{it}}}} = \theta _{{0{{i}}}} + \theta _{{{{1i}}}} {\text{FD}}_{{{{it}}}} + \theta _{{{{2i}}}} {\text{REM}}_{{{{it}}}} + \theta _{{{\text{3i}}}} {\text{REM}}_{{{{it}}}} *{\text{FD}}_{{{{it}}}} + \theta _{{{\text{4i}}}} {\text{X}}_{{{{it}}}} + \mu _{{{{it}}}}$$
(2)

where REMit*FDit denotes the interactive variable between financial sector development and remittances. This interaction term indicates how financial development affects the marginal effect of remittances on investment. The very total effect of remittances on investment is obtained as the sum of the direct effect and their indirect effect conditional on the financial development level:

$$\frac{{\partial I_{it} }}{{\partial {\text{REM}}_{it} }} = \theta_{2} + \theta_{3} {\text{FD}}_{it}$$
(3)

A positive (negative) coefficient on the interactive variable implies that the effect of remittances on investment increases (decreases) with financial development. On the other hand, Eq. (3) allows us to appraise whether, beyond a certain level, financial deepening fosters or reduces the effect of remittances on investment.

The common approach in many empirical studies testing for nonlinear effect is to simply look at the significance of the interaction coefficient θ3. The interaction variable will be dropped from the model if θ3 is found to be insignificant. However, basing inference on the coefficient θ3 may be misleading, as this approach ignores the covariance between θ2 and θ3. In addition, it is wrong to assume that for any value of financial development (FD), the marginal effect of remittances (REM) is significant. To make correct inference from the nonlinear model, we need to calculate the standard errors of marginal effects of remittances. Therefore, we compute these standard errors as follows:

$$\hat{\sigma }_{{\frac{\partial I}{{\partial REM}}}} = \sqrt {{\text{var}} (\theta_{2} ) + FD^{2} {\text{var}} (\theta_{3} ) + 2FD{\text{cov}} \left( {\theta_{2} ,\theta_{3} } \right)}$$
(4)

As can be seen, it is possible for the effect of remittances on investment to be significant for high values of financial development even if θ3 is insignificant.

Data description

This study uses annual data covering the period from 1985 to 2019 for 10 West African countries selected on the basis of data availability, namely, Benin, Burkina Faso, Cote d’Ivoire, Ghana, Mali, Niger, Nigeria, Senegal, Sierra Leone and Togo. Our dependent variable is domestic investment measured by total gross fixed capital formation scaled by GDP. Our variables of interest are remittances inflows as a share of GDP and financial development measured by domestic credit granted to the private sector expressed as a percentage of GDP. The financial indicator represents the activities of banks and other financial institutions in collecting savings. Remittances are defined as current transfers by migrant workers, and wages and salaries earned by nonresident workers. We further consider a set of other control variables including real GDP per capita in 2010 US dollars, trade openness approximated by the ratio of the sum of exports and imports to GDP, and inflation rate measured as the annual percentage change in consumer price index. The study also constructs an interactive term between financial development and remittances to measure the role of financial development in influencing the effect of remittances. All data were extracted from the 2021 World Development Indicators (WDI) of the World Bank. Table 2 provides information about the source, definition and measurement of the variables used in the empirical analysis.

Table 2 Variables used in this study
Full size table

Econometric methodology

Our econometric analysis is conducted as follows. We commence by testing for the order of integration of the variables by mean of panel unit root tests. Second, we test for the existence of a long-run relationship among the variables using panel cointegration tests. Third, we estimate the relationship between the variables. To that end, the relationship between financial development, remittances and investment was derived from a panel ARDL model specified as follows:

$$\begin{gathered} \Delta I_{it} = \alpha_{i} + \phi_{i} \left( {I_{it - 1} - \theta_{1i} FD_{it} - \theta_{2i} REM - \theta_{3i} REM_{it} \times FD_{it} - \theta_{4i} X_{it} } \right) + \sum\limits_{j = 1}^{m} {m_{0ij} \Delta I_{it - j} } + \hfill \\ \sum\limits_{j = 0}^{n} {m_{1ij} \Delta FD_{it - j} } + \sum\limits_{j = 0}^{p} {m_{2ij} \Delta REM_{it - j} } + \sum\limits_{j = 0}^{q} {m_{3ij} \Delta \left( {REM_{it - j} \times FD_{it - j} } \right)} + \sum\limits_{j = 0}^{r} {m_{4ij} \Delta X_{it - j} } + \mu_{it} \hfill \\ \end{gathered}$$
(5)

where Δ is the first difference operator, m1ij, m2ij, m3ij and m4ij are the short-run coefficients of financial development, remittances, interactive term and other control variables, θ1i, θ2i, θ3i, and θ4i are the long-run coefficients, and the coefficient ϕi shows the speed of adjustment of investment rate to covariates. The optimal lag structure (m, n, p, q, r) on the first-differenced variables are selected according to the Bayesian Information Criterion (BIC).

The ARDL approach was found to address the problem of endogeneity of some regressors. The potential endogeneity is deemed to arise, since remittances, real per capita GDP and financial development may be correlated with the dependent variable with possible reverse causality. The ARDL model mitigates the issue of endogeneity by incorporating the lag lengths for the exogenous and endogenous variables. The ARDL specification is more efficient for small sample and produces unbiased estimates of the long-run coefficients. Finally, the ARDL approach can be used where variables are integrated of different orders, such as I(0) and I(1) series.

The above error-correction model could be estimated using the Mean Group (MG) and Pooled Mean Group (PMG) estimators developed by Pesaran and Smith (1995) and Pesaran et al. (1999), respectively. The PMG and MG estimators are the two prominent estimators used in the empirical literature to analyze heterogeneous panels. The MG estimator allows both the short- and long-run coefficients to vary across countries. It estimates the model for each country and computes the estimates for the whole panel as simple averages of the individual coefficients. On the other hand, the PMG estimator as discussed by Pesaran et al. (1999), assumes homogeneity in the long-run coefficients (i.e., θ1i = θ1, θ2i = θ2, θ3i = θ3, θ4i = θ4) while allowing the short-run coefficients, the speed of adjustment, and the error variance to vary across countries. Under the assumption of homogeneity of the long-run coefficients, the PMG estimator is consistent and more efficient as compared to the MG estimator. The assumption of long-run homogeneity can be tested using the Hausman test.

One limitation of the PMG and MG approaches is that they ignore cross-sectional dependence when estimating the relationships among the variables. Results from these methods are inconsistent when cross-sectional dependence is present. In this study, we apply the CD test developed by Pesaran (2004) to ascertain that the long-run residuals are cross-sectionally independent.

Results and discussion

Descriptive statistics

The descriptive statistics and correlation matrix are shown in Table 3. As this table depicts, the representative country invested, on average, 18.885% of GDP over the period and reached its maximum at 54.948 and its minimum at − 2.424%. Real GDP per capita has an average of 6.680 and ranged between 5.609 and 7.844. Domestic credit to private sector has a mean value of 14.242% of GDP and reached its maximum at 41.156 and its minimum at 1.603%. Remittance inflows over GDP have an average value of 2.927% and a maximum value of 10.711% of GDP. Trade openness and inflation rate have a respective mean value of 56.091% of GDP and 10.083%. The standard deviations of the variables show that there is a great variability among countries. The Jarque–Bera test for normality rejects the null hypothesis of normality and suggests that all the variables are not normally distributed.

Table 3 Descriptive statistics and correlation matrix
Full size table

The analysis of the correlation matrix reveals a number of issues. First, the relationship between remittances and domestic investment is positive and significant, suggesting that remittances play a role in supporting domestic investment. Second, the relationship between financial development and domestic investment is positive but statistically insignificant. In addition, real GDP per capita has a positive and significant relationship with domestic investment. Finally, based on the correlation matrix, we observe that all the explanatory variables are not strongly correlated with each other as the coefficients correlation among them are weak.

Unit root and cointegration tests

The empirical analysis of this study begins with the linear specification, testing the separate effects of remittances and financial development on domestic investment. Next, we introduce the interaction term between remittances and financial development. Before proceeding with the regression estimates, we carry out some statistical tests in order to investigate the time series properties of the variables. More precisely, we first apply unit root tests to ascertain the order of integration of the variables. Then, we check the existence of a long-run relationship among the variables. The results of the IPS test by Im et al. (2003) and the ADF-Fisher test by Maddala and Wu (1999) are presented in Table 4. We also report the CADF test proposed by Pesaran (2007) which controls for cross-sectional dependence in the data. The results show that the dependent variable, investment rate, is non-stationary according to the three unit root tests and becomes stationary when first difference is considered. Similarly, real income, domestic credit and remittances have unit root in their levels but achieve stationarity when their first differences are considered. On the other hand, trade openness and inflation are stationary according to IPS and ADF-Fisher tests, but become non-stationary when cross-sectional dependence is controlled for. Overall, the results of unit root tests suggest that the dependent variable is I(1) while the explanatory variables are a mixture of I(0) and I(1) series. This justifies the use of the ARDL approach in modeling the relationship between the variables.

Table 4 Results of panel unit root tests
Full size table

After the confirmation of mixed stationary status of the variables, we examine the existence of a long-run relationship among them using Pedroni (2004) test. The results reported in Table 5 show that four of the seven tests support the existence of a long-run relationship among the variables at the 10% level of significance.

Table 5 Results of Pedroni panel cointegration tests
Full size table

Financial development and remittances improve investment

As cointegration exists among the variables, we estimate the long-run model using the PMG and MG estimators. The results are summarized in Table 6. The diagnostic tests show that the residuals from both PMG and MG estimators are cross-sectional independence and stationary. The Hausman test statistic clearly accepts the null hypothesis of homogeneity of the long-run coefficients. As discussed in the method section, under the null hypothesis of slope homogeneity, PMG performs better than MG. Based on the PMG estimates, we can observe that, except inflation rate, all the explanatory variables impact on investment rate significantly and positively. Therefore, expansion in the domestic credit to private sector ratio significantly increases domestic investment rate and this may spur economic growth of West African countries. More precisely, a one point of percentage increase in domestic credit to the private sector increases domestic investment rate by 0.142 point of percentage. This outcome supports the hypothesis H1 of the study. It also corroborates the supply leading hypothesis of Schumpeter (1911), Mckinnon (1973) and Shaw (1973) who observed that well-developed financial system spurs innovation and economic growth through funding of productive investment. Furthermore, our finding is consistent with Ndikumana (2000), Misati and Nyamongo (2011) and Yao (2018), but contradicts with Sakyi et al. (2016) and Iheonu et al. (2020) who reported insignificant effect of domestic credit to the private sector on domestic investment in Ghana and ECOWAS countries, respectively.

Table 6 Results of linear model
Full size table

Contrary to popular belief that remittances to developing countries are spent mainly on consumption and unproductive activities, our results show that remittances have favorable effects on domestic investment, implying that investment is among the motivations to remit. This finding is in line with the hypothesis H2 of the study. Some of the migrants may send remittances to home for investment purposes in perspective of their return. This finding aligns with Dash (2020), who also reported a positive effect of remittances on domestic investment. But our finding contradicts with the study of Iheonu et al. (2020) who found that remittances have insignificant effects on real gross fixed capital formation per capita in ECOWAS countries. In addition, our finding is not in line with Chami et al. (2005) who proved that remittances do not provide a source of capital formation for economic development. With respect to the other control variables, the results show that GDP per capita and trade openness positively and significantly impact on investment whereas the effect of inflation is insignificant.

Relationship between financial development and remittances: complementarity or substitutability?

We further investigate the complementarity or substitutability relationship between financial development and remittances in impacting on domestic investment. Table 7 presents the results from the nonlinear model with interactive effect between remittances and financial development. This table shows the mediating role financial development plays in the remittances and investment nexus. Consistent with our expectation, the interactive variable bears a positive and significant coefficient, which indicates that the effect of remittances on investment improves with the size of the financial sector. In other words, financial development has a complementary effect on the use of remittances for domestic investment. This finding supports the hypothesis H3 of the study, that is financial development enhances the positive effect of remittances on domestic investment in West Africa. The complementary hypothesis signifies that when financial system expands, remittances are used to save and invest in productive activities. When remittances pass through the financial system, they foster the supply of credit by providing liquidity to the market. The findings mirror those of Bettin and Zazzaro (2012), Bangake and Eggoh (2020), Ali Bare et al. (2022), and Nyeadi et al. (2022) who confirmed the complementary hypothesis between remittances and financial development in spurring investment and economic growth.

Table 7 Results of nonlinear model with interactive variable
Full size table

The negative coefficient on remittances and the positive coefficient on the joint effect suggest the existence of a threshold of financial development in the remittances–investment nexus in West African countries. This threshold level is estimated at 18.3%. Since the relationship between remittances and investment is finance-dependent, we compute the marginal effect of remittances across changes in domestic credit to private sector. Table 8 shows the marginal effects evaluated at the minimum, mean, median, maximum and threshold level of domestic credit to private sector. The results show that the marginal effect of remittances changes at different levels of financial development. It is negative at lower level of financial development and turns to positive at higher level of financial development after the threshold. For instance, at the minimum of domestic credit to the private sector, the marginal effect of remittances on investment rate is − 1.057 and statistically significant. This suggests that for a financial development level of 1.603% of GDP, a one point of percentage increase in remittances as a share of GDP will decrease domestic investment rate by 1.057 point of percentage. The increase in remittances will continue to reduce total investment rate until the threshold level of financial development is reached. At the threshold level (i.e., 18.3%), the marginal effect of remittances is zero, implying that any increase in remittances will not affect significantly domestic investment rate. Conversely, beyond the threshold level of financial development, any increase in remittance inflows will enhance domestic investment rate. For a financial development level representing 41.156% of GDP, a one point of percentage increase in remittance inflows enhances domestic investment rate by 1.453 point of percentage.

Table 8 Marginal effect of remittances on investment rate
Full size table

On average, for lower level of financial development, i.e. domestic credit lower than 18.3% of GDP, the relationship between remittances and investment is negative. All other things being equal, an increase of one point of percentage in remittances will decrease investment rate by 0.50 point of percentage. In financially underdeveloped countries, remittances occur through informal channels and are mainly spent on consumption, resulting in low saving and investment rates. Conversely, for higher level of financial development (i.e. domestic credit higher than 18.3% of GDP), the relationship between remittances and investment is positive. All other things being equal, an increase of one point of percentage in remittances as a share of GDP will enhance investment rate by 0.45 point of percentage. Thus, remittances spur investment rate in countries where the generosity of the financial sector is higher than 18.3% of GDP. The positive relationship between remittances and investment highlights the importance of remittances through development of the financial sector.

Is the relationship between remittances and investment significant for all levels of financial development? Fig. 3 illustrates how the marginal effect of remittances changes across the observed range of financial development. The 95% confidence interval around the line allows us to evaluate the significance of marginal effects. Remittances have a statistically significant effect on investment rate whenever the upper and lower bounds of the confidence interval are both above or below the zero line. It is easy to see that remittance inflows have a significant negative (positive) effect on investment rate when domestic credit granted to the private sector is less (greater) than 15% (21%). The marginal effect of remittances is not significant for levels of domestic credit to private sector ranging between 15 and 21% of GDP.

Fig. 3
figure 3

Marginal effect of remittances on domestic investment rate across financial development Note: The x-axis represents domestic credit to private sector as a share of GDP. The solid line represents the marginal effect of remittances on investment across domestic credit to private sector; the shaded grey area plots the 95% confidence interval for the marginal effect estimates

Full size image

Taken together, the results of this study provide evidence that remittances impact on domestic investment directly through increasing investment, and indirectly via increasing credit to the private sector. The study also reveals that remittances complement rather than substitute financial development in bringing about domestic investment. This is consistent with the complementarity hypothesis. The overall findings of this study support the three null hypotheses that we set out to probe. They suggest that the best way for SSA countries to ensure that remittances spur investment and economic growth is to increase remittance flows passing through the financial system at lower costs of transaction.

Conclusion and recommendations

The increased inflow of foreign remittances to developing countries and the recurrent financial crises around the world have attracted much interest among scholars about the growth effects of remittances and financial development. However, very limited work has been devoted to analyze the interactive effect of remittances and financial development on domestic investment and economic growth. From this background, this study sought to examine the relationship between remittances, financial development and domestic investment for a panel of 10 Sub-Saharan African countries over the period 1985–2019. The study goes beyond the direct effects of remittances and financial development on domestic investment by exploring the mediating role of financial development. This is important, because financial sector plays a significant role in channeling remittances and savings to investment. Moreover, differences in financial development can lead to differences in investment and economic growth.

The research adopts panel Pooled Mean Group estimator based on the results obtained from Hausman test. After establishing cointegration among the variables, we estimated the effect associated with each variable. The results from the linear model show that both remittances and credit granted to the private sector have positive and significant effects on domestic investment. When we estimate the nonlinear model, the combined effect of remittances and domestic credit to private sector was found to be positive. This evidence indicates that financial development enhances the effect of remittances on domestic investment. In other words, increasing credit to the private sector complements the positive role of remittances in enhancing capital formation. Thus, remittances can bring about more investment if financial sector is more developed. Other variables that positively drive investment are real GDP per capita and trade openness.

The results further indicate that there is a threshold effect of financial development in the remittances and investment relationship. Beyond a threshold of domestic credit to private sector estimated at 21% of GDP, there is a positive and significant relationship between remittances and investment, while this relationship is negative and significant under the threshold. Specifically, remittances are more likely to contribute to capital accumulation in countries with generous financial sector. Overall, the results are consistent with the supply leading hypothesis of the financial development theory advocated by Schumpeter (1911), Mckinnon (1973) and Shaw (1973) that well-developed financial system spurs innovation and economic growth through funding of productive investment.

Against the backdrop of the foregoing findings, this study recommends that West African countries should encourage financial institutions to increase the volume of credit to the private sector. The banking sector, which is the main source of credit to the private sector in many SSA countries, is an important channel of financial intermediation through which financial resources can be mobilized for investment. In these countries, banks refrain borrowing to private sector. Efforts should be done by governments towards increasing the confidence of banks via guarantee founds. Another recommendation of this study is that countries should implement policies that help increase official remittances. It is well-known that unrecorded remittances constitute a larger share of total remittance inflows to African countries. In addition, the costs of maintaining a bank account and the fees associated with loans are high in the region. On the other hand, the number of remittance recipients that access financial services in Sub-Saharan Africa is still small compared to the volume of remittances. Therefore, efforts should be focused towards increasing penetration and concentration of banks and promoting cost-effective payment technologies and inclusive services. Finally, West African countries can initiate investment vehicles such as diaspora bonds and “one migrant one account” operation to help finance their economies. The operation “one migrant one account” should allow migrant to make deposit into their saving accounts opened in their home country. All these policies may increase significantly the proportion of remittances that is saved and channeled to bring about higher economic growth.