Effects of International and Internal Remittanaces on Financial Inclusion in Ghana

This article examines the relationship between remittances and financial inclusion in Ghana. The data for the study was extracted from the results of an analytical review of the living standards survey indicators in Ghana. The methodological tools of the study are represented by a regression equation based on the use of the Force Entry Method to test the functioning of variables in the model. The study empirically confirms and theoretically proves that domestic remittances have a positive and significant impact on access to financial services, while international remittances affect the likelihood of opening a bank account, but do not have any significant impact on applying for a loan and lending to remittance households. It is substantiated that domestic and international money transfers have a significant positive impact on the opening of bank accounts, even when forging collateral. Based on the results of calculations, the paper substantiates the conclusion that remittances contribute to increasing the availability of financial services in Ghana. It was noted that domestic remittances have a greater potential to improve financial inclusion in Ghana than international remittances. The paper emphasizes that the provision of collateral is an important lever for lending to households. Remittances will have very little impact on financial inclusion when financial institutions require collateral to facilitate the application and grant. According to the results of the study, the following recommendation were provideds: development of a strategy to improve domestic remittances to increase indicators of financial inclusion and economic development; improving the conditions for remittances, especially domestic remittances, in order to ensure their flexibility and deepen financial integration; use of domestic remittances as collateral for household loans.


Introduction
Remittances are related to migration of people historically. In 2014, the World Bank reported that more than 200 million people are living in places other than their places of birth and this number continues to rise yearly. The migrants remit income to their home countries for various reasons including repayment of loans, funeral donations and establishment of small businesses for family members, contribution to family financial needs and personal investment (Aga &MertinezPeria, 2014). Developing countries in Africa, Asia and Latin America benefit a lot from remittances even though there are also outward remittances outside these countries to relatives in advanced economies mainly for the payment of school fees (Anyanwu & Erhijakpor,2010). They further explained that there is internal remittances where people remit money to support building projects, family support and investments though the recipients and senders of the remittance moneys live in the same country but in different locations. In most countries in Africa such as Kenya, Uganda, Ghana, Tanzania, Rwanda, mobile money applications have made internal remittances convenient, flexible and efficient. The sending and receipt of remittances has become a global concern over the years and researchers are interested in finding out the contribution of remittances to economic development in both micro and macro levels in host and home countries.
Demirguc-Kunt, et. al (2011) explain remittances as the cash and goods that are transferred to households by migrant workers who live outside the countries or communities they come from and are currently working in another country or community. They also explained that remittances may contribute to opening of bank accounts with financial institutions by the recipients. In 2010, the flow of remittances to African countries alone was US$40 billion making up 2.6% of Africa's GDP (World Bank, 2011). In 2017, remittances contributed more than US$2.2billion to the gross domestic product (GDP) of Ghana (World Bank, 2018). Though remittances contributed immensely to the economic growth and development of both home and host countries, (Aga &Mertinez Peria 2014) argued that the cost of remittance transfer and payment of commission to intermediaries may reduce the inherent benefits.
Financial inclusion has become an important area of studies when discussing migration and remittances. Studies have established that financial inclusion beneficial from remittances especially when various transfer applications are used including mobile money applications. It encourages effective distribution of funds (Ajefu & Ogebe, 2019) better access to finance, thus leading to faster correction of income disparity (Beck & Demirguc-Kunt, 2008; Honohan 2004), poverty reduction (Chibba, 2009). While these scholars believe that remittances positively influence financial inclusion, others believe remittances have no influence on financial inclusion. Anzoategui, Demirgüç-Kunt and Pería (2014) found a positive and significant impact of remittances on bank credit and bank deposit, while other empirical studies such as Guliano and Ruiz-Arranz (2009) found contrary evidence that remittances relax the financial constraint of the receiving household. This reduces the demand for loans from financial institutions with negative impact on financial inclusion. The two different schools of thought have raised the concern to examine whether or not remittances have any impact on financial inclusion in Ghana considering the migration rate of 48.6 % and annual remittance of GHȻ2.2billion contribution to GDP in 2017 (World bank ,2018). With the level of remittances flow in Ghana, there is the need to gain a proper understanding and clarity on the effect of remittances on financial inclusion. This study deviates from other existing studies on Ghana that examined only the impact of external remittances on economic growth.
The justification for Ghana as a study area is emphasized because there is limited and mostly inaccurate data on internal and external remittances, (Anzoategui et al., 2014). Githaiga and Kabiru (2014) advocated for more study on remittances and financial inclusion especially in developing countries where data is limited. This paper examines the extent to which internal and international remittances influence financial inclusion in Ghana.

Literature Review
Monahov, A. (2020) remittances have been reliable source of income and capital for households in small and emerging economies over many decades. The emergence of Covid-19 has therefore impaired liquidity stability for households in countries with underdeveloped financial markets. Financial inclusion is about using various tools such as technology and policies to integrate people into the formal financial services space irrespective of whether or not the recipients engage in economic activities. According to Mbutor and Uba (2013) financial inclusion is a strategy aimed at increasing the number of people in an economy who are banked and hence holding a formal bank account with banks and other formal financial institutions. Beck et al. (2006) also defined financial inclusion is a banking sector outreach that allows access to formal financial services and their use by households and organizations. In recent times financial inclusion encourages people to engage in formal financial system using mobile phones and the internet without necessarily holding a formal bank account. Thus financial inclusion aims at advancing the use of formal mode of payments, including cheques, ATM cards, internet payments, mobile payments and others by the populace. Financial inclusion is achieved by ensuring the ease of accessibility of financial services, its availability, and making use of formal financial system for all members of an economy (Shankar, 2013). Financial inclusion also serves as financial deepening due to its role in increasing the size of financial system, growing diversification of firms' and households' portfolios and developing the financial markets (Ajefu & Ogebe, 2019).
Mbutor and Uba (2013) explained that financial inclusion is the process that guarantees easy access, availability, and use of financial services for all participants in an economy. Embedded in this definition are the three main dimensions of financial inclusion-banking penetration (Bank accessibility), availability of banking services and the usage dimension.
Banking penetration is described by (Mbutor and Uba 2013) as the "the size of 'banked' population, that is, the proportion of people having a bank account. Alternatively, bank accessibility can be accomplished by counting the number of accounts that are opened through financial institutions and approximating the percentage of the citizenry with an account. Availability of banking services dimension measures the extent to which banking service is readily obtainable to people as and when it is needed. Gautam (2019). measured availability of banking service by "the number of bank outlets (per 1000 population) and/or by the number of ATM per 1000 people".
The dimension of usage represents the degree to which a person who holds an account with financial institution utilizes the banking services. Kempson et al. (2004) found that many people merely have accounts with financial institutions but barely utilize the service the financial institutions provide. Sarma (2008) measured the usage dimension as a proportion of the volume of credit and deposit to GDP. Githaiga (2014) and Nyamongo et al. (2009) concluded that remittances are a cross-border earnings that migrants send to their countries of origin. Remittances maybe cross regional or districts receipts within the same country when it is described as internal remittance.

Empirical review
Remittances and financial development

Theoretical review
Taylor (1999) distinguished between pessimistic and optimistic views underpinning remittances. Optimistic Views are rooted in Neo-Classical and Developmentalist theories. The Neo-Classical theory considers remittances flow as a means of optimizing resource distribution to benefit both the sending and the receiving countries. Inclusive redistribution of resources enhances the chance of economic growth (Todaro, 1969). "The theory, however states that an unconstrained movement of labor in a free market will lead to an increment in the marginal cost of production of labour that arises from the scarcity of labor in the migrant's sending country". The flow of remittance capital, is expected to move in an exactly opposite direction as migration of labour and hence leading to a factor price equalisation (the Heckscher-Ohlin model). The recipients of remittance money and how much to receive maybe defined by financial market tools that enhance participation hence financial inclusion.
Pessimistic views is rooted in the cumulative causation and the "Migrant Syndrome". Thus, the pessimists see migration as the major cause of disparity between the developed and the under-developed countries. Migrants are usually educated young men and women who easily get jobs. Remittances are sent to migrants' households, who are often better-off, tend to further deepen the income inequality in the migrants' country of origin (Lipton, 1980). The exposure of wealth of remittances receiving households has the propensity of changing the local taste of these households and leads to an increase in foreign demand (Lipton, 1980). This opposes the neoclassical view that remittances reduce poverty and inequality. Remittances maybe a continuous process that demands the use of financial services products and tools to effectively send or receive funds.

Model Specification and Data
This paper investigates whether international and internal remittances have impact on financial inclusion in Ghana. Financial inclusion includes bank accounts opening, loan request and loan grant. The dependent variables used in the model are dummy (Yes/No) in nature which include whether or not a household member has requested for a loan in the six months prior to the survey, whether or not the loan was granted and finally, whether a member of the household has a formal bank account with registered financial institution. The paper used binary logistic regression which models how binary response variables depend on a set of explanatory variables, which can be categorical, continuous or a mix of both of them in a model. The study used an annual cross-sectional data and the data was derived from Ghana Living Standard Survey round (GLSS6) which summarizes information on the living conditions and well-being of households in Ghana from the period 18th October 2012 to 17th October 2013 and has been in use until a new survey report was published. Out of 18,000 households, 16,772 households were successfully enumerated. The study does not change to GLSS7 because the paper has been produced and under review before the GLSS7 was released in October 2018. Remittances, as used in the study, comprise goods, cash and non-cash items received and issued by households The paper examines the impact of remittances on financial inclusion by considering:

Empirical Models
Using Binary logistic method, the paper employed Forced Entry Method, which by default allows all the predictor variables to be tested in one block to determine their predictive ability while controlling for the effects of other predictors in the model. The models adopt and modify certain household characteristics and human capital variables as used by Adams et al. (2008).

Financial Inclusion Model
The research seeks to empirically find whether remittances impact financial inclusion in Ghana. The paper adopted the modified model as used by Anzoategui, Demirgüç-Kunt, and Martínez Pería (2011). In their study, which is slightly close to this study, the choice of independent variables was based on education, gender and certain human characteristics. The dependent variable is represented by Y and x denotes the explanatory variable(s). In the linear regression it is assumed that the mean may be stated as an equation linear in x, such as: The probability of a household being financially inclusive can therefore be written as 1 * with * the latent response. Where; ExRemitRec is whether or not a member of each of the households receive remittances from abroad, InterRemitRec represents whether or not a member of each of the households receive remittances from other members within Ghana (internal or domestic remittances), ExterRemitGiv represents whether or not households members paid out remittances to others living outside Ghana, and finally, InterRemitRec represents whether or not households' members, from the survey results, paid out remittances to others living within Ghana.
HHSize represents the size of the households under consideration. That is the number of people living in each of the 16772 households. Males above18 represents the number of households that have male members who are 18years and above 18years, whereas Females above18 is the number households with female members who are 18 years and above, Salaries is a dummy variable that denotes whether or not households receive monthly salaries.
On the part of the human capital variables in the model, MembersSHS is the number of households with members who have attained SHS education, and MembersTer is the number of households that have members who have attained tertiary education.

Data
This section presents an initial summary of the variables being studied. Summary statistics comprising frequencies, means and standard deviations of the variables used in the model are discussed. Freq. represents frequency and Per. represents percentage.

Remittances and Loan Application
This sub-section provides findings based on the first objective of this paper-impact of remittances on loan application. Results are presented in Table 5. Logistic regression is used to examine the likelihood that a household will apply for a loan. Results presented include the coefficients (B), the p-values (Sig.) and the odds ratio (Exp (B)). Also reported is some goodness of fit indicators (Omnibus test and Hosmer and Lemeshow test) for the model. From table 5, the coefficient for received internal remittance is 0.233 with an odds ratio greater than 1. This indicates that as more households receive internal remittances, the chances that they will apply for loans increases. In relation to the odds ratio of 1.262, the implication is that households that received internal (domestic remittances) are 1.262 times more likely to apply for loans than households that do not receive remittances within Ghana. Again, households that paid remittances to others living within Ghana have positive impact on the request for loans, with an odds ratio of 1.79 greater than 1. This implies that as more households paid out remittances to people within Ghana, the higher will be the probability of loan application. Again , it is observed from Table 5 that households that have members who have attained tertiary education are 1.085 times more likely to request for loans from financial institutions than households without members with tertiary education. This is in line with the findings of Todaro (1976) and Shultz (1982) that as people pursue higher education they secure a better job and hence have the requirement for loan application from financial institutions. Again, households that have salaried workers are more likely to request for loans than households without salaried workers as members. This is due to the fact that the loan application process and requirement for salaried workers is less cumbersome

Remittances and Loan Grant
The second objective of this paper is to examine how remittances influence the acquisition of loans. Results are presented in Table 6. Logistic regression is used to examine the likelihood that a household loan will be granted. Results presented include the coefficients (B), the p-values (Sig.) and the odds ratio (Exp(B). Households that receive internal remittances are seen to significantly influence loan acquisition. The result shows an odds ratio of 1.465 and coefficient of 0.382 (Exp(B) = 1.465 , p < 0.001) supporting the view of Fajnzylber and Lopez (2008) that financial institutions are willing to give loans to households that they have enough information on as a result of frequent receipt of remittances. Households that paid out remittances to people within Ghana also have a significant impact on the acquisition of loans with an odds ratio of 2.069 more than households that do not receive internal remittances supporting previous studies (Anzoategui, 2014). Household size, though significant, negatively influences the probability of households receiving loans from financial institutions with odds ratio of less than 1 (0.901) indicating that the larger the household size, the less likely that the household will have their loans granted.
Certain household characteristics such as number of households that have female members who are 18 years and above, have significant impact on the probability that a household will receive a loan. This means that financial institutions are willing to grant loans to females than males. Again, the number of households with members having attained tertiary education has the highest odds ratio indicating a strong positively significant impact on loans acquisition. This confirms Mincer's (1974) assertion that education increases financial inclusion, specifically loan acquisition. Lastly, the result found a significant and positive relationship between loan grant and salaries. Households that have members who receive monthly salaries are almost twice likely to be granted loans than households whose members do not receive monthly salaries.  Table 7 shows the factors that interplayed in influencing financial institutions' decision to grant loans to some 1,954 households out of the 16,772 households that actually made a request for loans with financial institutions.
The focus of this test is to find out how remittances influenced the probability that financial institutions will grant loans to applicant households when loan application was indeed made. It is observed that only households that pay internal remittances 2.164, 0.001 households that have members with tertiary education and 1.368, 0.004 are the factors that are significant and positively influence applicants' loans to be granted. A household that sent remittances within the country had 2.164 times likelihood to receive a loan from a bank than households that did not send remittances to others. Again, households that have members with tertiary education also had 1.368 chances of receiving loans from financial than households that do not have members with tertiary education. The result affirms some of the earlier finding in table 6 that financial institutions are willing to grant loans to applicant with tertiary level of education because of the minimum risk of default among such members. Households that have members who receive monthly salaries actually were twice more than households than households that do not receive any monthly salaries in terms of loans grants.

Remittances and Bank Account Opening
In this sub-section, direct logistic regression was performed to examine how remittances influence the chances of households opening a bank account with financial institutions, the third objective of this paper, on the part of households that engage in remittances. For simplicity purpose, and as done in the previous sub-sections, the coefficients (B), the p-values (Sig.) and the odds ratio (Exp(B) are reported in Table 8. The result in Table 8 shows that households that received internal remittances have a high chance of opening accounts with financial institutions. The odds ratio of 1.413 indicates that households that received remittances from Ghana are over 1.4 times more likely to open an account than households that do not receive internal remittances. This is so because many of the internal transfers are often done with financial institutions.
Again, the result shows that households that sent remittances to others living outside Ghana are 2.841 times likely to open a bank account than other households that do not send remittances abroad. This explains that households that send remittances abroad will need to have an account with financial institutions in order to enjoy certain benefits like ATM transaction, online banking that may facilitate smooth transfer of remittances to people living outside Ghana.
Certain household characteristics in the model such household size, number of households that have female members who are 18 years and above significantly but negatively influence the probability that a household will have an official bank account. As expected, the paper found that households with members who are salaried workers are likely to have an official account with financial institutions. This is the case because most monthly salaries are paid through financial institutions and hence the likely tendency that such members will have an official account.
On education, we found that as members of households graduate from one level of education to the other (From SHS to tertiary institutions) the probability of the household opening an account increases accordingly. We also found that the number of households with members having tertiary education, which is the strongest predictor, is 8 times more likely to open an account than households that do not have members with tertiary level education. This indicates that, in the model and among the variables used, households with more educated members are more likely to have a bank account, most especially tertiary education.

Control for Collateral
The paper further takes into consideration assets that are owned by households which can serve as collateral for loans request and loans grant. Assets that serve as collateral in this paper include home appliances, lands and buildings and vehicles that are owned by the 16772 households.  Table 9 presents the coefficients (B), the p-values (Sig.) and the odds ratio (Exp(B). We found that land and building owned by households is positively significant to loan application with odds ratio and P-Value of 1.200, 0.009 respectively. In essence, households that possess land and building are 1.2 times more likely to request for loans than households that do not own land and building. Household size has odds ratio of 0.113, 0.000 indicating that household size is significant but negatively related to loan request. Households with members that have attained tertiary education are also more likely to apply for loans with odds ratio and P-Values of 1.085, 0.022 .   Table X, land and building is significant and positively correlated with loan grant with odds ratio and P-Value of 1.324, 0.043 respectively. Households that own land and building are 1.324 times more likely to be granted loans than households that do not have land and building. Other variables that are significant and positively impact the chances of a household being granted a loan are households with female members who are 18 years and above. Again, households whose members have attained SHS and tertiary education are likely to be granted loans with odds ratio and 1.114, 0.000 respectively. Household size, however, reduces the chances that a household will be granted loans. This may be due to the high dependency ratio that exists in households in Ghana specifically the 16,772 households under consideration.  Again, households are 1.352 times more likely to open an account to enhance the transactions than households that do not send remittances abroad. In essence a unit increase in the number of households that sends remittances abroad increases the probability that households will open an account with formal financial institutions by 1.045.
Human capital used in the model in the form of formal education shows that education has a higher tendency on improving bank account opening among households as found in the results in Table 11 when collateral was not factored. Again, variables in the model that represent households' characteristics (age in years of household heads, number of female members in a household who are 18 years and above and household size) is significant but negatively influence the probability of a household opening account with financial institutions. Households that have land and buildings have a positive probability of opening an account with an odds ratio and P-value of 1.996, 0.002 respectively. These households are likely to have an account with financial institutions, among other reasons, to increase their chances of acquiring loans when the need arises

Diagnostics for the results
The study employed two main diagnostics to assess the adequacy of all the results obtained. The omnibus test assesses the overall fitness of the model and thus it is based on the null hypothesis that there is no significant joint effect of the independent variables on the dependent variable. The p-value of the omnibus test in all the regression results led to the rejection of the null hypotheses and thus the conclusion is that all the models estimated are fit. Hosmer-Lemeshow test assesses the goodness of fit test of a regression model, specifically a probit or logit regression. The null hypothesis of the Hosmer-Lemeshow test is that there is a "non-poor" fit. Thus, it could be observed from all the results that all the regression estimations are fit.

Conclusion and Recommendations
This paper empirically investigated the extent to which both internal and international remittances impact financial inclusion in Ghana. It is concluded that internal remittance significantly influences financial inclusion in Ghana. Internal remittances positively and significantly influence financial inclusion in all the three objectives (loan request, loan grant and bank account).
On external remittances, the paper concludes that external remittances do not significantly influence financial inclusion in Ghana. Specifically, external remittances do not to have any significant impact on loan application and loan grant but have a positive and a significant impact on bank account opening.
On ownership of assets, the paper concludes that a household with land and building increases the probability of applying for loan from financial institutions. Thus, collateral strengthens access to household credit. Physical asset collaterals weigh more than expected remittances in loan application success.
Internal remittances have higher potential to improve financial inclusion in Ghana than international remittances. Remittances will have a very little impact on financial inclusion when financial institutions demand collateral security to foster loan application and grant.

Recommendations
Strategies to improve internal remittances is recommended to improve financial inclusion and deepen economic development Enhance environment for remittance especially internal remittance to make it flexible deepen financial inclusion.
Use internal remittances as collateral for loan by households.

Acknowledgment
Mr. David Korsah is acknowledged for helping in data extraction. Mr. Mac Abeka Junior is also acknowledged for making valuable comments.