The Effects of Remittances on Financial Development in Africa

(1)

The Effects of Remittances on Financial

Development in Africa

Master thesis in Finance by A.J. Mensink, s2373416 June 13, 2016

Supervisor: dr. J. Bolt

Faculty of Economics and Business University of Groningen

ABSTRACT

This thesis examines the relatively unexplored relationship between remittances and financial sector development. More specifically, the effects of remittances on credit to the private sector and gross domestic savings are being studied. For this study a panel of 54 African countries over the period 1980-2011 is used. The results show that remittances have no impact on credit to the private sector and a negative impact on gross domestic savings. These findings are therefore not in line with the view that remittances promote financial development.

Keywords: Remittances, Financial development, Credit to the Private Sector, Gross Domestic

(2)

2

1. Introduction

Remittances, funds received from migrants working abroad, have increased enormously since the beginning of 2000. At the end of 2011, the inflow of remittances to African countries was the most important source of income received from abroad. According to Figure 1, remittances were even higher than foreign direct investments (FDI) and official development assistance (ODA) received in 2011. The effectiveness of FDI and ODA inflows on the econ omy of developing countries has always been a favored subject in the research area of development finance. Remittances, however, hardly received any attention by researchers in the previous century. Nowadays, it is more than relevant to understand the effects of remittances on reducing poverty and increasing economic growth due to the rapid increase of remittances and the characteristics of these inflows. Firstly, the rapid increase can be attributed to an increase in globalization and mobilization of labor, but also to lower transaction costs of remitting income. It is assumed that remittances are even higher than presented in Figure 1, because a sizeable part of remittances is received through informal channels, which are unknown to formal banks and money transfer companies. Secondly, the characteristics of remittances differ from the other international capital inflows. Remittances provide a stable income for the relatives of the migrant in the home country and are seen as less volatile. In addition, remittances are on average countercyclical, because migrants tend to increase their remittances when an economic shock occurs in their home country.

Figure 1: Capital inflows to African countries (in millions of US$).

Note: Vertical axes; millions of US$, Horizontal axes; year between 1980 and 2011. FDI refers to foreign direct investment. ODA is the official development assistance. Remittances are workers’ remittances, compensation of employees and migrants’ transfers (World Bank, 2016).

(3)

3

One of the key drivers in increasing economic growth and reducing poverty is financial sector development (Khan et al., 2000; Hassan et al., 2011). The main functions of the financial sector include mobilizing savings, facilitating the trade of goods and services and efficiently allocating resources (Zhuang et al., 2009). While the development of the financial sector could occur in numerous ways, this study focuses on the size of credit and savings of an economy. The higher the amount of credit and savings, the more developed a country is in terms of financial depth. When financial sector development is of such importance in achieving long-term economic growth and lowering poverty levels, it is interesting to study the impact of remittances on financial sector development.

Previous literature does not show consensus on the effects of remittances on financial sector development, as arguments exist for both positive and negative effects of remittances on financial sector development. Firstly, when remittances are received through banks, this will give banks the opportunity to offer new products to its clients (Orozco and Fedewa, 2005). In addition, individuals without a bank account could be reached more easily, which increases the scope of the financial sector. Furthermore, as receivers of remittances are characterized by their relative stable income, less risk is associated with these households. These are attractive clients for banks, so that credit could be granted at lower interest rates (Aggarwal et al., 2010). A final reason for a positive effect is that remittances will increase the loanable funds that a bank could offer (Coulibaly et al., 2015). The main reason why remittances will induce higher savings is the increasing need for products that provide storage of these remittances (Aggarwal et al., 2010).

However, this evidence is not conclusive, because arguments for a negative effect exist as well. Firstly, remittances could function as a substitute for credit by relaxing financing constraints (Aggarwal et al, 2010). In that case, credit from banks is no longer required, because remittances will function as a form of funding. This will even discourage to make use of the financial sector. Secondly, individuals will not increase their savings when remittances are directly used for consumption purposes. Especially for African countries there could be distrust in the financial institutions, which will not stimulate making use of the banking sector (Aryeetey, 2005). This thesis contributes to a relatively new debate on the impact of remittances on financial development1. Most research that has been done stresses the importance of remittances on increasing financial development. For example, Aggarwal et al. (2010) argue that this

1_{Financial development and financial sector development have the same meaning in this thesis. It refers in this}

(4)

4

relationship is positive for developing countries. However, existing research focuses on moderate financially developed countries, which increases the possible existence of endogeneity. In the case of endogeneity, remittances do not only stimulate financial development, but also vice versa. By focusing on less financially developed countries, it is worthwhile to examine whether remittances still contribute to financial development or if they may discourage financial development, taking into account the arguments mentioned in the previous paragraph. Moreover, this study contributes to existing literature by studying both the impact on credit to the private sector, as well as gross domestic savings. Credit to the private sector is an important measure of financial depth and therefore an indicator of financial development (King and Levine, 1993). Gross domestic savings is an indicator of financial deepening and therefore also assumed to be important in financial development (Mohan, 2006). An increase in gross domestic savings will increase the availability of funds for investment purposes, and thereby encourage financial sector development.

To study the impact of remittances on these measures of financial development, two models are constructed including a number of control variables that also relate to financial development. Firstly, a random effects model with credit to the private sector as the dependent variable. Secondly, a fixed effects model with gross domestic savings as the dependent variable. These models are however, based on the assumption that the time series are stationary. Therefore this problem is addressed in a separate section by changing the variables with an unit root into first differences in order to end up with two stationary models. This method is then applied to a sample of all 54 African countries for the period 1980-2011.

The results show that remittances have a negligible impact on credit to the private sector and a negative and significant impact on gross domestic savings. These findings indicate that remittances are not stimulating financial development in terms of these two measures, which contradicts the popular view of Aggarwal et al. (2010) in which remittances do enforce financial development.

(5)

5

2. Literature review

(6)

6

would be more productive investments and a better allocation of capital. Less diverse results are found on the relationship between remittances and poverty. Adams (2005) provides strong empirical evidence for a reduction in poverty, in terms of level, depth and severity due to an increase in both international migration and remittances for developing countries. Remittances are especially successful in reducing poverty for the poorest households (Adams., 2006). For these households the remittances meant a dramatic change of their income which diminished the poverty severely. One of the main reasons why remittances are successful in decreasing poverty is the stabilizing impact of remittances on household income (Gupta et al., 2009).

Although remittances seem to be effective in reducing poverty, it increases the potential for differences in income distribution (Adams, 2009; Anyanwu, 2011). Anyanwu et al. (2011) show empirical evidence for an increase in income inequality after an increase in remittances, where income inequality is measured by the Gini coefficient. This supports the idea that receivers of remittances are on average wealthier and an increase in remittances will change their income status even more. This leads to differences in spending habits and even more diversification in income levels is expected.

There is also growing literature on the effects of remittances on spending behavior. Several views exist on the way remittances are spent on consumption, savings and investment. The first view indicates that there exist no differences between the use of remittances and other income. There also occurs no change in marginal spending behavior according to Adams et al. (2008), who investigated the effects of remittances on spending behavior in Ghana. No differences are found in the amount of money spent on investment and consumption goods between households with and without remittances. A second view highlights the negative effects of remittances in which these inflows are to a large extent used for “status-oriented” consumption goods (Chami et al., 2005). Only a small proportion is used for saving and investment purposes, but overall the remittances tend not to stimulate the productivity. This is line with the study of by Azam and Gubert (2006), in which they argue that in case of an adverse shock in the home country, remittances are mainly used for consumption. Again, this refers to the stabilizing impact of remittances. A third view states that remittances actually do increase private investments. For Sub-Saharan Africa, Githaiga (2014) did observe a positive impact of remittances on private investments, where these inflows provide a good source of capital.

(7)

7

development of the financial sector. Until now, a theoretical relationship is not established between remittances and financial development. Some researchers argue that remittances stimulate financial development, whereas others state that remittances function as a substitute. In the latter case, the financial sector is circumvented and remittances are directly used for financing purposes.

One strand of literature advocates a positive relationship between remittances and financial development. The findings of the research of Gupta et al. (2009), Giuliano et al. (2009) and Nyamongo et al. (2012) indicate that remittances do promote financial development. It seems that remittances are effective in reducing liquidity constraints, which improves the access to credit. In addition, remittances tend to increase financial investments in the home country. For developing countries, Aggarwal et al. (2010) concluded that remittances lead to an increase in both deposits and credit to the private sector as a percentage of GDP, both measures of financial depth. Furthermore, there is evidence that remittances lead to an increase in the use of financial services, both formal and informal. Ambrosisus and Cuecueha (2016) found statistical significant effects of remittances on various variables like the ownership of saving accounts and the existence of debt. While the majority of previous research argues that remittances stimulate financial development, there are also criticizers proclaiming that remittances are substitutes of financial development. In less financially developed countries, in terms of the size of the banking sector, remittances could function as an alternative way of financing investments (Guiliano et al., 2009). In this way credit markets are avoided, hence discouraging financial development. Instead, remittances are directly used to better allocate capital, which leads to higher levels of economic growth. Considering domestic savings, Hossain et al. (2014) observed even a negative impact on domestic savings explained by the crowding-out effect of remittances. That is, remittances tend to displace domestic savings. An argument for this could be the lower pressure to save for the home countries, because they expect remittances to increase in bad economic times.

(8)

8

In addition remittances even have a negative impact on gross domestic savings. Where Aggarwal et al. showed a positive effect of remittances on deposits, the results of this study agree with the view of Hossain et al. (2014) on the effects of remittances. This means that remittances tend to partly displace savings.

3. Hypotheses

The two measures of financial development used in this study are credit to the private sector and gross domestic savings. The former refers to financial resources provided to the private sector, whereas the latter is calculated as GDP minus consumption expenditure (Appendix A.2.). Based on the literature review, it is not clear a priori how these two variables will respond to an increase in remittances. Following the results of Aggarwal et al. (2010), however, two hypotheses will be stated about the expected outcomes. The first hypothesis deals with the credit to the private sector, which is generally used as a measure for financial depth. It is expected that an increase in remittances will cause an increase in potential clients and a boost in the offering of new financial products and services to its clients. Furthermore, the loanable funds will increase and clients with remittances will be favored by banks due their relative stable income. Therefore, the following is expected:

Hypothesis 1: Remittances will have a positive impact on credit to the private sector.

The second hypothesis considers gross domestic savings. While Aggarwal et al. (2010) researched the impact on deposits, the same line of reasoning could be used in the expectations of remittances on gross domestic savings. That is, the remittances could provide funds for investing purposes and therefore these remittances will stimulate gross domestic savings. This would be the desired effect, because financial development will increase. But when remittances are mainly used for consumption there will be no evidence for the following hypothesis.

Hypothesis 2: Remittances have a positive impact on gross domestic savings.

(9)

9

4. Methodology

The effects of remittances on financial development are studied using an unbalanced panel with 54 African countries over the period 1980-2011. Aggarwal et al. (2010) were the first authors that specified a relationship between measures of financial development and remittances in a regression analysis. Their model is therefore adopted in testing the two hypotheses of this research. In this paper, the main goal is to identify the effects of remittances on credit to the private sector and gross domestic savings. Credit to the private sector is seen as an important determinant in financial development and economic growth (King and Levine, 1993; Aggarwal et al., 2010). Due to the lack of data availability, gross domestic savings is used as a measure instead of deposits, which would be a better indicator of financial sector development. Even though this will decrease a direct comparability between the results of Aggarwal and this paper, the relationship still makes sense. According to Mohan (2006), gross domestic savings is a measure for financial deepening and therefore, it is assumed to play an important role in financial development.

The dependent variable of interest in this research is the remittances as a percentage of GDP. Remittances are defined as: “current transfers by migrant workers and wages and salaries of nonresident workers” (World Bank, 2016). Beside remittances there are five control variables included which have a relationship with the dependent variables according to the research of Aggarwal et al. (2010).

The first control variable is the log of GDP. A higher GDP means that the fixed costs regarding financial development become less important. The second variable is the GDP per capita, which is a measure of the development of the economy. A higher GDP per capita also means on average more legal institutions (Aggarwal et al., 2010). The third control variable is inflation, which is assumed to be negatively correlated. Higher inflation lowers the macroeconomic performance of a country and as a result, it negatively impacts financial development. The fourth and fifth control variables according to the research of Aggarwal et al. (2010) deal with the degree of current and capital openness. One is the exports to GDP and the other, FDI inflows to GDP. The more open a country is in terms of its exports and FDI inflows, the more a country is financially well developed.

(10)

10

data availability. These include portfolio inflows, aid inflows and a dummy for exchange rate regime. When these omitted variables are correlated with other independent variables, this could lead to an omitted variable bias. This thesis will continue by assuming that this bias will not play a significant role. The second issue relates to the so called reverse causality, where financial development stimulates remittances, because of lower transaction costs involved in remitting. According to the research of Coulibaly (2015), financial development does not seem to determine the amount of remittances received in Sub-Saharan African countries. This reduces the concern for reverse causality in this research. A final bias related to endogeneity is the measurement error, due to the fact that a high number of remittances are received through informal channels (Luna Martinez, 2005). Therefore remittances are expected to be underestimated, which decreases the reliability of the results.

In order to analyze the panel data obtained for this research, two different techniques are used to regress the measures of financial development on the independent variables. For credit to the private sector, the random effects model is more consistent according to the Hausman test (Appendix A.4.). While the fixed effects model is more appropriate for gross domestic savings. The individual specific effect is thus uncorrelated with the explanatory variables in the former and correlated in the latter.

Furthermore, some regression diagnostics are used in order to assess the validity of the model. The first issue deals with non-normality of the data, even the residuals are not normally distributed. A reason for this could be the very extreme outcomes for some variables. However, the results of statistical testing can still be assumed to be valid, because the sample size is large enough (Ghasemi and Zahediasl, 2012). Furthermore, heteroscedasticity and serial correlation are found to be present in both models. These issues are partly solved by using robust standard errors for testing the hypotheses. That is, errors are heteroscedasticity and autocorrelation consistent, which provide more reliable standard errors. In addition multicollinearity is not an issue for this dataset, because the highest correlation found is 0.38 (Appendix A.8.). For simplicity, it is now assumed that all variables are stationary. In a later stadium of this research the problem of non-stationarity will be addressed. For now, research continues with the two equations provided below.

𝐶𝑟𝑒𝑑𝑖𝑡𝑖,𝑡= 𝛽1𝑅𝑒𝑚𝑖,𝑡+ 𝛽2𝐿𝑜𝑔𝐺𝐷𝑃𝑖,𝑡+ 𝛽3𝐿𝑜𝑔𝐺𝐷𝑃𝑐𝑎𝑝𝑖,𝑡+ 𝛽4𝐼𝑛𝑓𝑖,𝑡+ 𝛽5𝐸𝑥𝑝𝑖,𝑡+ 𝛽6𝐹𝐷𝐼𝑖,𝑡+ 𝜔𝑖,𝑡 (𝑬. 𝟏) 𝑤ℎ𝑒𝑟𝑒 𝜔𝑖,𝑡= 𝜀𝑖,𝑡+ 𝑣𝑖,𝑡

(11)

11

The subscript 𝑖 refers to the country and 𝑡 refers to the year of interest. 𝐶𝑟𝑒𝑑𝑖𝑡 and 𝑆𝑎𝑣𝑖𝑛𝑔𝑠 refer to the measures of financial development, namely credit to the private sector and gross domestic savings. Remittances, 𝑅𝑒𝑚, is the important independent variable and in addition control variables are incorporated in the equations. Finally, the 𝜔 and 𝑢 are the error terms of the random effects model and the fixed effects model respectively. These two equations will be used in evaluating the impact of remittances on financial development.

5. Data

The models developed in the previous section are performed over all 54 African countries. In total there are eight variables included in Equation 1 and Equation 2. These variables are all World Development Indicators gathered from the World Bank database. This contains annual data over the period between 1980 and 2011. The year 1980 is used as a starting point, because the quality in the period before is poor and hardly available. The data availability of the used period is still limited, so there is an unbalanced panel. Two general limitations should be mentioned here. First, exact data is hard to obtain for the African countries, especially for remittances. Not all remittances are received through formal channels, such as banks and money transfer companies. This means that actual remittances are higher than the observed number. Second, there is quite some variability in the results as can be seen below from the large standard deviations for most variables in Table 1. This table provides summary statistics of the dependent and independent variables for the period 1980 till 2011. Data is not complete for each time period and country, which explains the differences in number of observations.

Table 1: Descriptive statistics.

Variable Observations Mean Standard _deviation Minimum Maximum

Credit (% GDP) 1492 20.31 20.38 0.68 167.54

Savings (% GDP) 1461 10.18 19.87 -103.42 89.63

Remittances (% GDP) 1210 4.20 10.71 0.00 106.48

Log of GDP (in constant US$) 1561 9.49 0.66 7.88 11.29 GDP per capita

(in thousands US$) 1561 1.01 1.47 0.05 9.23

Inflation (% of GDP) 1571 47.22 710.71 -33.13 26762.02

Exports (% of GDP) 1552 30.36 18.65 1.95 101.75

FDI (% GDP) 1538 3.44 9.91 -82.89 145.20

Note: Credit (% GDP) refers to credit to the private sector in % of GDP. Savings (% GDP) refers to gross domestic savings in % of GDP.

(12)

12

highest average credit ratio to GDP of 110.2%. The gross domestic savings is on average 10.18% of GDP. There are quite some countries with a negative savings rate, which is most likely due to high consumption. Lesotho shows the lowest gross domestic saving rate of -54.8%. Whereas Equatorial Guinea exhibits the highest average of 50.4%. Remittances as a percentage of GDP are between 0% and 106.5% according to Table 1, with an average of 4.20%. Although the top receiving countries of remittances in absolute amounts are countries outside Africa, quite some countries show a high dependency to remittances as a percentage of GDP. Figure 2 shows the top receiving countries in terms of remittances to GDP. Lesotho is the largest recipient with remittances of 60.78% of GDP. This could be due to the fact that Lesotho is a small economy, so that remittances to GDP are relatively high. Other countries with a high dependence on remittances are Cape Verde (15.70%), Liberia (9.32%), Egypt (7.25%) and Morocco (6.90%). Summary statistics on the control variables are also presented in Table 1. It should be noted here that the high maximum of inflation refers to the hyperinflation in Congo in 1994. Because the outlier does not seem to affect the results and assumptions, this observation remains in the sample. In the end, there was no reason for excluding observations from the dataset.

Figure 2: Top 10 receivers of remittances in Africa (in % of GDP).

(13)

13

6. Empirical results

6.1. Estimation Results

Table 2 presents the main outcomes of Equation 1 and Equation 2 for the sample of African countries for the period 1980-2011. Beside remittances, all control variables are listed on the left hand side. The impact on both credit to the private sector and gross domestic savings, as a percentage of GDP, are presented in Table 2. It appears that both variables are negatively correlated with remittances, although the coefficient for remittances in the first equation is not found to be significant.

Table 2: Results Equations 1 & 2.

Variables Credit to private sector (% GDP) Gross domestic savings (% GDP)

Remittances (% GDP) -0.08 -0.50***

(0.06) (0.06)

Log GDP (in constant US$) 8.05*** 6.62

(2.14) (4.88)

GDP Per Capita

(in thousands US$) 7.15*** -0.61

(0.91) (1.90) Inflation (% GDP) -0.00 0.00*** (0.00) (0.00) Exports to GDP (% GDP) -0.16*** 0.36*** (0.04) (0.07) FDI to GDP (% GDP) 0.04 0.10* (0.06) (0.05) Constant -57.83*** -61.79 (19.91) (45.19) Observations 1,114 1,084 R-squared 0.11 0.34 Number of countries 51 49

Note: Robust standard errors in parentheses; *Statistical significances at the 10% level **Statistical significances at the 5% level *** Statistical significances at the 1% level.

(14)

14

The results show that hypothesis 2 does not hold either. The sign of the coefficient for remittances is negative and significant. More specifically, an increase of one percentage point in remittances leads to a decrease of 0.50 percentage point in gross domestic savings to GDP. A priori, one could argue that savings would increase due to the remittances received for investing purposes in the future. The opposite holds however, which tends to increase consumption and it is possible that their savings are used for consumption as well. This is in line with the crowding-out argument of Hossain et al. (2014). That is, part of the existing savings tend to decrease, because the need to maintain high saving rates is lower. The control variables appear to have the expected impact, at least for the random effects model for credit to the private sector. An increase in both GDP and GDP per capita has a positive and significant impact on credit as a percentage of GDP. This is in line with the results of Aggarwal et al. (2010), where financial development is enhanced by a higher level of GDP and higher GDP per capita. That is apparently due to relatively lower fixed costs and an increase in legal institutions. Besides, the exports to GDP are also found to be negative and significant, which is the opposite of what would be expected. A recent study of David et al. (2014) on the link between trade openness and financial development in Africa could partly explain this outcome. These researchers examined that this causal relationship holds for countries that already have a high degree of institutional quality. Countries in Africa are in general poorer in terms of financial development and higher exports do therefore not seem to have a positive impact on the financial development. It appears, however, that when unit roots are taken into account, the model gives different outcomes, so less weight should be put on this result.

(15)

15

6.2. First differences model

The research continues by introducing two new equations. Until now, it was assumed that all variables are stationary. Because this research deals with time series variables, the existence of unit roots could be a potential bias. Therefore the augmented Dickey-Fuller test is performed, which reveals that some variables contain unit roots (Appendix A.9. till A.11.). The variables that are non-stationary are credit to the private sector to GDP, log of GDP and GDP per capita. Non-stationary data can lead to a spurious regression in which there is a relationship between non stationary variables. This could exist, because most variables, both dependent and independent, are related to GDP. And this will give bias in the testing procedure. The problem of stationarity is addressed by the following two regressions:

∆𝐶𝑟𝑒𝑑𝑖𝑡𝑖,𝑡= 𝛽1𝑅𝑒𝑚𝑖,𝑡+ ∆𝛽2𝐿𝑜𝑔𝐺𝐷𝑃𝑖,𝑡+ ∆𝛽3𝐿𝑜𝑔𝐺𝐷𝑃𝑐𝑎𝑝𝑖,𝑡+ 𝛽4𝐼𝑛𝑓𝑖,𝑡+ 𝛽5𝐸𝑥𝑝𝑖,𝑡+ 𝛽6𝐹𝐷𝐼𝑖,𝑡+ 𝜔𝑖,𝑡 (𝑬. 𝟑) 𝑤ℎ𝑒𝑟𝑒 𝜔𝑖,𝑡= 𝜀𝑖,𝑡+ 𝑣𝑖,𝑡

𝑆𝑎𝑣𝑖𝑛𝑔𝑠𝑖,𝑡= 𝛽1𝑅𝑒𝑚𝑖,𝑡+ ∆𝛽2𝐿𝑜𝑔𝐺𝐷𝑃𝑖,𝑡+ ∆𝛽3𝐿𝑜𝑔𝐺𝐷𝑃𝑐𝑎𝑝𝑖,𝑡+ 𝛽4𝐼𝑛𝑓𝑖,𝑡+ 𝛽5𝐸𝑥𝑝𝑖,𝑡+ 𝛽6𝐹𝐷𝐼𝑖,𝑡+ 𝑢𝑖,𝑡 (𝑬. 𝟒)

Table 3: Results Equations 3 & 4.

Variables ∆Credit to private sector (% GDP) Gross domestic savings (% GDP)

Remittances (% GDP) -0.00 -0.52***

(0.01) (0.06)

Log GDP (in constant US$) 0.40 1.62

(4.32) (1.53)

GDP Per Capita (in thousands US$) 4.33* 1.11**

(2.40) (0.51) Inflation (% GDP) -0.00** 0.00*** (0.00) (0.00) Exports to GDP (% GDP) 0.02** 0.39*** (0.01) (0.07) FDI to GDP (% GDP) 0.00 0.12** (0.01) (0.06) Constant -0.33 0.38 (0.25) (2.20) Observations 1,100 1,078 R-squared 0.11 0.33 Number of countries 51 49

Note: Robust standard errors in parentheses; *Statistical significances at the 10% level **Statistical significances at the 5% level *** Statistical significances at the 1% level.

(16)

16

not seem to influence the amount of credit to the private sector. This result is still in conflict the first hypothesis. The same holds for the second hypothesis, the negative impact has even strengthened a little. An increase of one percentage point in remittances lead on average to a decrease of 0.52 percentage point in gross domestic savings as a percentage of GDP. Furthermore, the first difference of the log of GDP is not found to be significant in both models. In a sample with more countries with a higher GDP, it is perhaps more likely to draw stronger conclusions on the impact of GDP on financial development. On the other hand, GDP per capita has a positive and significant impact on both dependent variables. A possible reason for this is that the amount of credit and savings increase with institutional quality. Additionally, the inflation coefficient for the credit model turns out to be negative and significant in Equation 3. This result was also found for developing countries by Aggarwal et al. (2010), where inflation tends to worsen macroeconomic performance. Hence, it is less likely that credit is granted to the private sector. Moreover, the sign of exports to GDP is changed by introducing the first differences. An increase of one percentage point increases the (first difference of) credit to the private sector as a percentage of GDP by 0.02 percentage point. The impact is small, but significant. Thus, it seems that trade openness indeed is

beneficial for financial development.

The first differences model provides better estimates, but the main conclusion from Table 3 is in line with Table 2. That is, the effect of remittances on credit to the private sector is negligible, but negative and highly significant on gross domestic savings.

6.3. Different time periods

(17)

17

2000 the decrease is lower (0.43 percentage point). This could indicate a small relative increase in financial deepening, although the crowding-out effect still seems to be present.

Table 4: Different time periods.

1980-1999 2000-2011 Variables ∆Credit Savings ∆Credit Savings

Remittances (% GDP) 0.00 -0.72*** -0.04 -0.43***

(0.01) (0.05) (0.04) (0.09)

Log GDP

(in constant US$) -0.71 1.13 4.23 -0.05

(4.19) (1.10) (3.61) (1.08)

GDP Per Capita

(in thousands US$) 1.91 0.96** 2.20 1.31***

(1.21) (0.41) (1.60) (0.40) Inflation (% GDP) -0.00** 0.04* -0.06** 0.03 (0.00) (0.02) (0.03) (0.04) Exports to GDP (% GDP) 0.02** 0.47*** 0.03* 0.42*** (0.01) (0.07) (0.02) (0.10) FDI to GDP (% GDP) -0.01 0.01 0.06** 0.14* (0.01) (0.05) (0.03) (0.08) Constant -0.43 -1.09 -0.35 -2.21 (0.33) (2.04) (0.62) (2.98) Observations 653 654 443 420 R-squared 0.22 0.37 0.25 0.26 Number of countries 46 44 45 43

Note: ∆Credit refers to ∆Credit to private sector (% GDP). Savings is defined as Gross domestic savings (% GDP). Robust standard errors in parentheses; *Statistical significances at the 10% level **Statistical significances at the 5% level *** Statistical significances at the 1% level.

7. Conclusions

The main goal of this thesis was to investigate the effects of remittances on financial development of countries in Africa. Firstly, no relationship has been found between remittances and credit to the private sector, which indicates that remittances are not associated with an increase in financial depth. Credit to the private sector has been an interesting measure for financial development in previous studies.

(18)

18

remittances stimulate financial development, by using different measures. It remains therefore an interesting subject for future research.

Secondly, the results indicate a negative and significant impact of remittances on gross domestic savings. This indicates that consumption is relatively higher and savings relatively lower, which is in line with the research of Hossain et al. (2014). The possible effects of remittances are then twofold. Firstly, it appears that the extra income, remittances, is not used for saving purposes. And secondly, there even seems to exist a partly crowding-out effect of savings in which remittances are substituting for savings. An argument for this could be that relatives in the home country rely on remittances in poor economic times. Therefore it is plausible that receiving households are assured of a relatively stable income so that the pressure to save is lower. As previously mentioned, there are some limitations to this thesis. On the one hand there are some data issues and on the other hand methodological concerns. With respect to the data, the non-normality of the variables and the presence of serial correlation and heteroscedasticity could, despite the robust standard errors, have biased the results of the testing procedure. With regards to the methodology, an important limitation could be the possible existence of endogeneity due to a number of reasons. The first reason could be the omitted variable bias. More control variables could have been added to the models. Besides exports and FDI inflows, more variables can control for current and capital openness. These could be aid inflows, portfolio inflows or a dummy variable for exchange rate regime. They are present in the methodology of Aggarwal et al. (2010), but excluded in this thesis because the data was hard to obtain for African countries. Adding these variables would increase the reliability of the results. A second reason is the possible presence of reverse causality. As Aggarwal et al. (2010) explained in his study, financial development decreases the transaction costs and therefore stimulates to send remittances. For the simple reason that African countries are financially not well developed, it was assumed not to play an important role in this research. A third reason for possible endogeneity is measurement error. The data on remittances do not reflect all the remittances received by households. This is mainly because remittances via informal channels are not registered (Luna Martinez, 2005). Remittances are underestimated, hence decreases the strength of the conclusions that are made.

(19)

19

(20)

20

8. References

Adams, R. H., & Page, J. (2005). Do international migration and remittances reduce poverty in developing countries?. World development, 33(10), 1645-1669.

Adams, R. H. (2006). Remittances and poverty in Ghana (Vol. 3838). World Bank Publications.

Adams Jr, R. H., Cuecuecha, A., & Page, J. (2008). Remittances, consumption and investment in Ghana. World Bank Policy Research Working Paper Series.

Adams, R. H. (2009). The determinants of international remittances in developing countries. World Development, 37(1), 93-103.

Aggarwal, R., Demirgüç-Kunt, A., & Pería, M. S. M. (2010). Do remittances promote financial development?. Journal of Development Economics, 96(2), 255-264.

Ambrosius, C., & Cuecuecha, A. (2016). Remittances and the Use of Formal and Informal Financial Services. World Development, 77, 80-98.

Anyanwu, J. C. (2011). International remittances and income inequality in Africa. Review of Economic and Business Studies, 7, 117-148.

Aryeetey, E. (2005). Informal finance for private sector development in Sub-Saharan Africa. ESR Review, 7(1), 13.

Azam, J. P., & Gubert, F. (2006). Migrants' remittances and the household in Africa: a review of evidence. Journal of African Economies, 15(suppl 2), 426-462.

Chami, R., Fullenkamp, C., & Jahjah, S. (2005) Are immigrant remittance flows a source of capital for development?. IMF staff papers, 52(1), 55-81.

Coulibaly, D. (2015). Remittances and financial development in Sub-Saharan African countries: A system approach. Economic Modelling, 45, 249-258.

David, A. C., Mlachila, M., & Moheeput, A. (2014). Does Openness Matter for Financial Development in Africa?. IMF Working paper.

Ghasemi, A., & Zahediasl, S. (2012). Normality tests for statistical analysis: a guide for non-statisticians. International journal of endocrinology and metabolism, 10(2), 486-489.

Githaiga, P. N. (2014). Do Remittances Stimulate Private Sector Investment? A Case of Sub-Saharan Africa. European Journal of Business and Management, 6, 29.

Giuliano, P., & Ruiz-Arranz, M. (2009). Remittances, financial development, and growth. Journal of Development Economics, 90(1), 144-152.

(21)

21

Hassan, M. K., Sanchez, B., & Yu, J. S. (2011). Financial development and economic growth: New evidence from panel data. The Quarterly Review of economics and finance, 51(1), 88-104.

Hossain, D. (2014). Differential Impacts of Foreign Capital and Remittance Inflows on Domestic Savings in Developing Countries: A Dynamic Heterogeneous Panel

Analysis. Economic Record, 90(s1), 102-126.

Kaberuka, D., Namubiru, R. (2014). The effects of remittances on gross domestic savings in Uganda (1999-2011). International Journal of Business Management and Administration, 3(2), 29-39.

Khan, M. M. S., & Semlali, M. A. S. (2000). Financial development and economic growth: an overview (No. 0-209). International Monetary Fund.

King, R. G., & Levine, R. (1993). Finance and growth: Schumpeter might be right. The quarterly journal of economics, 717-737.

Lucas, R. E., & Stark, O. (1985). Motivations to remit: Evidence from Botswana. The Journal of Political Economy, 901-918.

de Luna-Martinez, J. (2005). Workers' remittances to developing countries: a survey with central banks on selected public policy issues. World Bank Publications.

Mohan, R. (2006). Economic Growth, Financial Deepening and Financial Inclusion, Address at the Annual Bankers' Conference 2006, Hyderabad on November 3. In Migrant Worker Remittances, Micro-finance and the Informal Economy: Prospects and Issues, Working Paper No. 21, Social Finance Unit, International Labour Office.

Nyamongo, E. M., Misati, R. N., Kipyegon, L., & Ndirangu, L. (2012). Remittances, financial development and economic growth in Africa. Journal of Economics and Business, 64(3), 240-260.

Orozco, M., Bump, M., Fedewa, R., & Sienkiewicz, K. (2005). Diasporas, Development and Transnational integration: Ghanaians in the US, UK and Germany. Institute for the Study of International Migration and Inter-American Dialogue.

Zhuang, J., Gunatilake, H. M., Niimi, Y., Khan, M. E., Jiang, Y., Hasan, R., ... & Huang, B. (2009). Financial sector development, economic growth, and poverty reduction: A literature review. Asian Development Bank Economics Working Paper Series, (173).

Data

Africa Development Indicators. (n.d.). Retrieved March 12, 2016, from http://data.worldbank.org/data-catalog/africa-development-indicators

Migration and Remittances Data. (n.d.). Retrieved March 12, 2016, from

(22)

22

Appendix

A.1. List of African countries

Algeria Libya Angola Madagascar Benin Malawi Botswana Mali Burkina Faso Mauritania Burundi Mauritius Cameroon Morocco Cape Verde Mozambique Central African Republic Namibia

Chad Niger

Comoros Nigeria Congo, Dem. Rep. Rwanda

Congo, Rep. Sao Tome and Principe Cote d'Ivoire Senegal

Djibouti Seychelles Egypt, Arab Rep. Sierra Leone Equatorial Guinea Somalia Eritrea South Africa Ethiopia South Sudan Gabon Sudan Gambia, The Swaziland Ghana Tanzania Guinea Togo Guinea-Bissau Tunisia Kenya Uganda Lesotho Zambia Liberia Zimbabwe

A.2. Variable definitions

Source: Africa Development Indicators. (n.d.). Retrieved March 12, 2016, from

http://data.worldbank.org/data-catalog/africa-development-indicators

Variable Short definition Source

Domestic credit to private sector (% of GDP)

Domestic credit to private sector refers to financial resources provided to the private sector, such as through loans, purchases of nonequity securities, and trade credits and other accounts receivable, that establish a claim for repayment. For some countries these claims include credit to public enterprises.

International Monetary Fund, International Financial Statistics and data files, and World Bank and OECD GDP estimates.

Gross domestic savings (% of GDP)

Gross domestic savings are calculated as GDP less final consumption expenditure (total consumption).

(23)

23 Workers' remittances and compensation of employees, received (% of GDP)

Workers' remittances and compensation of employees comprise current transfers by migrant workers and wages and salaries earned by nonresident workers. Data are the sum of three items defined in the fifth edition of the IMF's Balance of Payments Manual: workers'

remittances, compensation of employees, and migrants' transfers.

World Bank staff estimates based on IMF balance of payments data, and World Bank and OECD GDP estimates.

GDP (constant 2000 US$)

GDP at purchaser's prices is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. Data are in constant 2000 U.S. dollars. Dollar figures for GDP are converted from domestic currencies using 2000 official exchange rates. For a few countries where the official exchange rate does not reflect the rate effectively applied to actual foreign exchange transactions, an alternative conversion factor is used.

World Bank national accounts data, and OECD National Accounts data files.

GDP per capita (constant 2000 US$)

GDP per capita is gross domestic product divided by midyear population. GDP is the sum of gross value added by all resident producers in the economy plus any product taxes and minus any subsidies not included in the value of the products. It is calculated without making deductions for depreciation of fabricated assets or for depletion and degradation of natural resources. Data are in constant U.S. dollars.

Inflation, GDP deflator (annual %)

Inflation as measured by the annual growth rate of the GDP implicit deflator shows the rate of price change in the economy as a whole. The GDP implicit deflator is the ratio of GDP in current local currency to GDP in constant local currency.

Exports of goods and services (% of GDP)

Exports of goods and services represent the value of all goods and other market services provided to the rest of the world. They include the value of merchandise, freight, insurance, transport, travel, royalties, license fees, and other services, such as communication, construction, financial, information, business, personal, and government services. They exclude compensation of employees and investment income (formerly called factor services) and transfer payments.

Foreign direct investment, net inflows (% of GDP)

Foreign direct investment are the net inflows of

investment to acquire a lasting management interest (10 percent or more of voting stock) in an enterprise

operating in an economy other than that of the investor. It is the sum of equity capital, reinvestment of earnings, other long-term capital, and short-term capital as shown in the balance of payments.

(24)

24 Prob>chi2 = 0.2355

= 8.04

chi2(6) = (b-B)'[(V_b-V_B)^(-1)](b-B) Test: Ho: difference in coefficients not systematic

B = inconsistent under Ha, efficient under Ho; obtained from xtreg b = consistent under Ho and Ha; obtained from xtreg FDI .0445366 .042096 .0024405 .0121497 EXP -.1610233 -.1612274 .0002041 .0102166 INF -.0017506 -.0016232 -.0001273 .0002923 GDPCAPTHO 8.690287 7.153245 1.537042 .6728412 LOGGDP 6.066106 8.052066 -1.98596 1.634835 REM -.0928195 -.0803648 -.0124547 .0152026 fixed random Difference S.E.

(b) (B) (b-B) sqrt(diag(V_b-V_B)) Coefficients

A.3. Hausman test for Equation 1 (Random effects model)

A.4. Hausman test for Equation 2 (Fixed effects model)

(V_b-V_B is not positive definite) Prob>chi2 = 0.0000

= 33.88

chi2(6) = (b-B)'[(V_b-V_B)^(-1)](b-B) Test: Ho: difference in coefficients not systematic

B = inconsistent under Ha, efficient under Ho; obtained from xtreg b = consistent under Ho and Ha; obtained from xtreg FDI .1022972 .0678968 .0344004 .0073416 EXP .3634194 .3661888 -.0027694 .0080295 INF .0030369 .0033117 -.0002748 .0001205 GDPCAPTHO -.6124535 .1887464 -.8011999 .5259264 LOGGDP 6.61835 8.082073 -1.463723 1.132759 REM -.495836 -.5153994 .0195634 .0101263 fixed random Difference S.E.

(25)

25 Prob > chi2 = 0.0000

chi2(6) = 1045.12

Variables: REM EXP INF LOGGDP FDI GDPCAPTHO Ho: Constant variance

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity A.5. Normality test for all variables

A.6. Breusch-Pagan test for heteroscedasticity (Random effects model)

A.7. Breusch-Pagan test for heteroscedasticity (Fixed effects model)

A.8. Correlation matrix (test for multicollinearity)

CREDIT SAV REM LOGGDP GDPCAP INF EXP FDI CREDIT 1 SAV 0.15 1 REM -0.04 -0.63 1 LOGGDP 0.50 0.52 -0.24 1 GDPCAPTHO 0.50 0.51 -0.11 0.43 1 INF -0.05 0.06 -0.02 -0.01 -0.03 1 EXP 0.13 0.39 -0.02 0.03 0.50 0.08 1 FDI -0.05 -0.10 0.10 -0.11 -0.03 -0.00 0.19 1 SAV 1461 0.90586 83.875 11.135 0.00000 FDI 1538 0.37600 582.632 16.039 0.00000 GDPCAPTHO 1561 0.61102 368.139 14.891 0.00000 LOGGDP 1561 0.98163 17.384 7.197 0.00000 INF 1571 0.03013 923.266 17.213 0.00000 EXP 1552 0.92279 72.686 10.800 0.00000 REM 1210 0.36247 478.361 15.398 0.00000 CREDIT 1492 0.71020 263.200 14.021 0.00000 Variable Obs W V z Prob>z

Prob > chi2 = 0.0000 chi2(6) = 521.37

Variables: REM EXP INF LOGGDP FDI GDPCAPTHO Ho: Constant variance

(26)

26

A.9. Fisher-type unit-root test (test for stationarity) for ‘Credit to the private sector’

A.10. Fisher-type unit-root test (test for stationarity) for ‘Log GDP’

A.11. Fisher-type unit-root test (test for stationarity) for ‘Log GDP’

Other statistics are suitable for finite or infinite number of panels.

P statistic requires number of panels to be finite.

Modified inv. chi-squared Pm -0.6117 0.7296

Inverse logit t(264) L* 1.5267 0.9360 Inverse normal Z 1.4181 0.9219 Inverse chi-squared(104) P 95.1776 0.7202 Statistic p-value Drift term: Not included ADF regressions: 0 lags

Time trend: Not included Panel means: Included

AR parameter: Panel-specific Asymptotics: T -> Infinity Ha: At least one panel is stationary Avg. number of periods = 27.79 Ho: All panels contain unit roots Number of panels = 52

The Effects of Remittances on Financial Development in Africa