This section describes the variables that are estimated in the regression model, presents the regression model and also how the data is collected.
Previous Hypotheses and Their Testing
Since our field of studies is relatively new we have not been able to find any previous study on our hypothesis. However, we have looked at how previous researchers have dealt with our variables. Here we present the result for the two major variables; microcredit and remittances.
The need to measure low-income households’ access to microcredit has increased however there is a lack of such data. It is difficult to find data on the number of micro-enterprises that are borrowing, who gets access to financial services and what services are most needed in low-income areas (Honohan, 2005). It is difficult to statistically measure the impacts of microcredit; one way could be to measure whether average income has increased for borrowers that have received microcredit (Armendàriz de Aghion & Morduch, 2005).
Mosley and Hulme (1998) have studied the impact of 13 poverty-reducing, microfinance institutions on financial performance and income in seven countries. The hypothesis in their regression is that microfinance instruments reduce poverty. Their result is that microfinance programs are barely efficient to help households that are very poor, since these households need to use the borrowed capital for basic needs. Consequently, the regression result show on an upward sloping curve where the poorest households are negatively influenced by a loan. For households with some type of starting capital, however, a loan can help to start up a business and the loan can be repaid, thus the impact of a loan is positive.
Studies have showed that selection bias is common in microcredit programs, those who already have some sort of capital are more likely to join or have access to a program. Thus the effect is larger than it would be if all individuals would apply (Armendàriz de Aghion & Morduch, 2005). Coleman (2006) has studied microcredit in Thailand to approximate for the impact it has for poor and at the same time control for the selection bias. Many microcredit programs goal is to reach those who are the poorest among the poor. However the selection bias work in two ways; first is that villagers that chose to participate in a program and receive a loan are more likely to have entrepreneurial skills from the beginning. They are also more likely to be able to create a higher income for themselves already before they start which makes comparisons of program members and nonmembers biased. The other way is that programs are more likely to be placed in rural areas that are perceived as more organized, entrepreneurial and better governed than others. These two factors imply that the villagers would have a higher welfare than others even if the program did not exist.
To collect data Coleman (2006) made surveys that included 444 household in 14 different villages in Thailand. The result of the regression is that the microcredit programs fail at their attempt to reach the poorest of the poor. The households that are a part of the programs are those who already before the start are wealthier than the poorest.
Giuliano and Ruiz-Arranz (2006) have studied remittances in about 100 developing countries. Their study tests how much remittances matter for growth in a country, dependent on how well the country is developed financially. Giuliano and Ruiz-Arranzs (2006) data comes from the International Monetary Funds, (IMF), Balance of Payments Statistics Yearbook and the different countries authorities. They define remittances as transfers of money made by migrants to their country of origin.
Their regression results for the first regression are attached in Appendix 2. The results from the first regression are that remittances have almost no impact on growth when financial development is not taken into account. However if investments are not counted for, remittances has some impact on growth.
Giuliano and Ruiz-Arranz (2006) second regression regress the relationship between remittances and growth when financial depth is involved. The hypothesis is to test what effect remittances has on growth when the financial depth in the country is counted for. Therefore a variable of financial development (FinDev) added to the model. The second regression results are attached in Appendix 3. The findings are that remittances’ impact on growth decrease as the level of financial development increase. This implies that financial needs can be met by the banking system when it is functioning and vice versa; remittances play more of a financial role when there is no functioning bank system (Giuliano & Riuz-Arranz, 2006).
A Revised Hypothesis
A model of variables that is important in developing countries for individuals to be able to start a new firm has been estimated. The regression analysis was conducted by the selection method of backward elimination. One of the limitations in the paper is the fact that microcredit is not based on a specific theory so the regression is based on the theoretical research that was conducted. The authors have not been able to find that this kind of study has been previously made. Therefore the regression trails have been exploratory in the sense that different ways of arranging the data has been tried and different variables have been used before the result was achieved. The thesis is built upon Equation 7:
Y = a + b1 Micro + b2 Rem + b3Corr + b4 Edu + b5 Infra + b6 Pop + b7 GFCF + e ; (7)
In Table 4.1 the independent variables of the regression model are defined and in Table 4.2 the hypothesis for the independent variables are explained.
The field of study is relatively new and when one works with developing countries data is hard to access. The authors have been in contact with several agencies that deal with development and microcredit to get help with data. The data that is most difficult to access is the number of micro, small and medium enterprises in low income countries. For example the Swedish International Development Cooperation Agency (Sida) was interested in such data but did not know where to find it. Due to the fact that it is difficult to retrieve data from low -income countries lower-middle -income countries have been added to the data to have a large enough dataset. Appendix 4 lists all included countries. There are two control variables and those are; population growth and gross fixed capital formation in current US$.
The data for the dependent variable (Y) comes from the World Development Indicators (2008). The dependent variable is represented by the number of micro, small and medium enterprises in the countries. As standard definition micro enterprises have up to 9 employees, small enterprises have up to 49 employees and medium enterprises have up to 249 employees. It was either this or using data for new firm entries as the dependent variable. The reason micro, small and medium enterprises was chosen is that the authors find that variable more interesting since the firm type that is discussed in this thesis belongs to that category. The problem with new enterprise data is that it contains information about registered business and in poor countries only 20-30 percent of employment outside of agriculture belongs to this category thus most microenterprises operate without registration.
The microcredit variable data is closely related to access to commercial banks. In this data all types of institutions that provide microcredit (banks, NGOs, non-profit organizations and banks) are included. The Micro data is the number of active borrowers of micro institutions in each country which are reported to MIXmarket (2008). The remittances variable data is retrieved from the World Development Indicators (2008). The Rem data is workers’ remittances and compensation of employees, from Balance of Payment in current USD. The difference between remittances and compensation in this context is that remittances come from people who have been migrant workers for more than one year, while compensation is from people who have migrated from the country for less than one year. The corruption variable data was taken from Transparency International (2008) The Corr variable is measured by a Corruption Perception Index which measures to what degree politicians and public officials are perceived as corrupt. Businesses and experts answer surveys about their perception of the degree of corruption and the results from that is put together in a composite index Transparency International (2008).
The infrastructure variable data is retrieved from the World Development Indicators (2008). The Infra data is percentage of paved roads in relation to the country’s total road-network. A problem with this kind of data is that a country can have many roads, but if they are not paved their data numbers become small. For reasons that are explained in the theory section, one should not put emphasis on the result of this variable in the regression analysis. One reason not to include more variables to explain infra is that it might cause multicollinearity. Multicollinearity is when there is a relationship between all or some of the independent variables in a regression model (Gujurati, 2003). The education variable was retrieved from the World Development Indicators (2008) and is measured by the number of students at the secondary level of education.
Result and analysis
Microcredit, Remittances, Corruption, Infrastructure, Population Growth, Gross Fixed Capital Formation and Education where regressed on the number of micro, small and medium size firms and the results are reported in Table 5.1.The regression was estimated by Ordinary Least Squared (OLS) and is for the time period of 1997 to 2005 and the data is pooled. These years were chosen because of available data for all the variables. 38 low- and low-middle income countries with available data for these years where included and tested on a = 0.1. The F-statistics is 4,504 and the R-squared is 0,233. The constant, Micro-variable, Rem-variable, Corr-variable and Edu -variable are significant ata = 0.1 and all except the Rem-variable have the right signs according to our hypothesis. The Infra-variable, Pop-variable and GFCF-variable are not significant at a = 0.1 and the Infra-variable and the GFCF-variable have a negative sign, which is not in line with the hypothesis.
The F-statistics is 4,504 and from Gujurati (2003) one can find a statistical table that reports F-distribution, for this regression the degrees of freedom are 6 in the numerator and 31 in the denominator which is equal to a F -distribution of 1,98 at a = 0.1 . 4,504>1,98 thus one can conclude that at least one of the independent variables are not equal to zero and thus has a relationship to the dependent variable. There are more variables than the ones included in the regression that affect the phenomenon that the regression seeks to explain. The R² is a measure of goodness of fit of the regression, the R² from the regression reported in Table 5.1 is 0,233. Thus about 25 percent of the variations in the number of micro, small and medium size firms in low and lower-middle income countries are explained by the regression model.
The control variables are added to the model to be able to show a more accurate level of the dependent variable. The Pop-variable has the expected positive sign and a significance of 0,246; however one cannot conclude that it is true since it is not significant at the 10 percent level. If the Pop-variable increases with one unit, the dependent variable (Y) increases with 874 763 units. The GFCF-variable has a negative sign and a significance of 0,344, although one cannot conclude that this negative relationship holds. If the GFCF-variable increases with one unit, the dependent variable (Y) decreases with 0,00001 units.
The Micro-variable has a positive sign as expected and a significance of 0,001, thus one can say that access to microcredit has a positive impact on micro, small and medium enterprises in developing countries. If the Micro-variable increases with one unit, the dependent variable (Y) increases with 1,795 units. The Rem-variable has a negative sign and a significance of 0,024; however the impact of this variable is very low, close to zero. If the Rem-variable increases with one unit, the dependent variable decrease with 0,0002 units. One reason for this might be that lower-middle -income countries are included and they are expected to have a more developed financial sector. According to Giuliano and Ruiz-Arranz (2006) remittances affect countries with a more developed financial sector less. Another reason for the negative sign can be that as people receive an increased amount of remittances, they might start to relay on that money, and decrease their own efforts to earn money. This can also be linked to the introduction where the problem of aid is discussed, where the problem of moral hazard exists; when people know that they will receive money, they will most likely make less effort to change their own situation. One can see the same relationship in the backward supply for labor; when an individual earns a higher wage, he is inclined to work less.
The Corr-variable has a negative sign as expected and a significance of 0,098. Corruption seems to have a very negative impact on micro, small and medium firms which can be explained by the relatively lower income retrieved when a small business owner has to use a fraction of his income to pay bribes. When a country is corrupt it takes longer time to start a business and fewer entrepreneurs will be able to start up an own business. If the Corr-variable increases with one unit, the dependent variable decreases with 1 183 052 units.
The Edu -variable has a positive sign as expected and a significance of 0,033. It is as in line with the hypothesis and previous research, that entrepreneurs are positively related to the education level. If the Edu-variable increases with one unit, the dependent variable (Y) increases with 0,206 units.
The Infra-variable has a negative sign and a significance of 0,456. The negative sign is not in line with the hypothesis; however since it is not significant one cannot conclude that this relationship holds. If the Infra -variable increases with one unit, the dependent variable decreases with 3,562 units. This achievement could be explained by the fact that percentage of paved roads is the only variable included as infrastructure and this might not be sufficient. The percentage of paved roads depends upon the size of the country and also how large the network of roads for that country is, thus it might be misleading.
Conclusions and Suggestions for Further Studies
The purpose of this paper was to test if the economic growth in developing countries, measured by firm existence, is influenced by access to microcredit, remittances, corruption and infrastructure. The empirical analysis suggests that microcredit; remittances and corruption have a significant effect on the number of micro, small and medium enterprises operating in developing countries. This can also be seen when one study the bank situation in Sweden in the nineteen century one can see that people benefit from the growth of local banks. The conclusion of the Swedish bank example is to see the connection between Sweden as a developing country and the developing countries today. In the twenty first century, a way to spread banks locally and improve welfare could be through microcredit. However, one cannot conclude that the infrastructure variable has an impact on the dependent variable, this could be due to the lack and poor quality of data for developing countries that has been discussed. In reality one can see from previous studies of Sweden for example, that infrastructure has a strong positive impact on firms, thus one should be hesitant to draw any conclusions from the results of infrastructure in the regression.
Arthur Lewis (1954) theory of labor moving from the agricultural to the industrial sector could be enhanced by microcredit. The regression result shows that by microcredit, individuals can create their own jobs. Therefore, they can advance from surplus labor to become a part of the working society and perhaps in the long run create jobs for others that are surplus labor.
Since the results show that microcredit affects the number of micro, small and medium enterprises in a country one can conclude that it can be used as a way out of poverty. Study of previous literature combined with the regression results can also conclude that microcredit is probably used most efficiently when it is given to individuals with some years of education and are not the poorest of the poor. In contrast to Mosley and Hulme’s (1998) result, which showed that microfinance is barely efficient to help poor households, the regression model suggests that microfinance is significantly positive and do help people in developing countries. This difference can be explained by the included low-middle income countries that have a higher income level, who can more easily take advantage of microcredit programs. By further studies one can exclude these countries and only focus on low income countries, which could give another result.
Bohman’s (2008) findings of poor regions that have even distribution of income is in line with Mosley and Hulme’s (1998) argument that when individuals are too poor, microcredit is not effective. This is due to the fact that all of the region’s income goes to basic needs; however when there is uneven distribution in a poor region some individuals might afford to engage in microcredit, by starting firms. If some individuals in a region start firms it is most likely beneficial for the whole region.
Further studies on the effect of microcredit can be made measured by different development indicators such as electricity consumption. If the consumption of electricity has increased since microcredit was accessible one can assume that microcredit has improved the development in the country. One can also make a field study to access more data, especially on our dependent variable; the number of micro, small and medium enterprises. Also political stability, taxes, subsidies and information are variables that could be included in a regression model that measures the number of micro, small and medium enterprises in a country.
As previously stated, individuals that have a low or non-existing capital stock cannot undertake investments that might be profitable if they do not have access to capital from some source. One of these resources necessary can be remittances. In short, remittances is when emigrants send money home to their families or when an individual migrate and work to save up money that he then brings home. The link between remittances and small and medium enterprises is shown in Figure 2.1 on page 2. If an individual continuously receives remittances it can be seen as a sort of security; there is inflow of a certain amount of capital. Woodruff & Zenteno (2001) has studied remittances to Mexico, where most of the people who migrate come from low income rural areas. The migration to receive a higher wage is not necessarily an emigration to another country it can also be a migration within the country to urban areas, however, most Mexicans immigrate to the US (Woodruff & Zenteno, 2001). According to geographical economics there is a spatial wage structure, which shows that there is a negative relationship between distance from an economic center of activity and wages. In the rural areas the distance is often large; hence there is an incentive to move to urban areas (Brakman, Garretsen & Van Marrewijk, 2001). Woodruff and Zenteno (2001) found that remittances are a relatively large share of the initial capital for microenterprises in Mexico. The small businesses are often started in urban areas, even by those who live in rural areas, due to a larger access to the market.
Lucas (1987) has studied labor migration to South African mines from other African countries. He found that wages in the mines had to be higher than wages individuals could earn at home, to create incentives to migrate. The effects of such migration might differ depending on the existents of surplus labor. If there is surplus labor in the country where people migrate from, the economy is not affected in any major way from migration. However, if there is not surplus labor, the effects in the home economy might be that wages increase due to lack of laborers, which in the long run can induce those who migrated to return home. The decrease of laborers might even in the short run lead to a decrease in output in the home country, however Lucas (1987) finds that in the long run remittances that are used for capital investment actually increase the agricultural output.
Giuliano and Ruiz-Arranz (2006) have found that remittances role is enhanced when the financial level is less developed in a country. This is because it can play the financial part of investments, like financing the start-up and survival of new firms. In this context remittances can almost be seen as an alternative or substitute to microcredit, because people in developing countries with poor credit markets will rely more on money from their relatives which does not require collateral or high repayment costs. Even if remittances are more important in countries that are less developed financially, in reality most remittances actually flow in to countries that are more developed financially. The reason for this might be that there is reduced transfer cost and it is easier to yield a higher return from investments when there are developed financial markets in an economy (Giuliano & Ruiz-Arranzs, 2006).
According to Giuliano and Ruiz-Arranz (2006) research of developing countries, remittances are the second largest mean of external financing, larger than aid. Wimaladharma, Pearce and Stanton (2004) has found the same relationship, where Foreign Direct Investment, (FDI) is the only capital inflow that is larger. Giuliano and Ruiz-Arranz (2006) have found that money made abroad and either sent home or saved and brought home are in a large part used to finance the start-up of microenterprises. Dustmann and Kirchkamp (2001) have studied the behavior of migrations both as they leave their household and when they return. They have found that re-migrates are often involved in entrepreneurial economic activity when they return.
Orozco and Fedewa (2006) argue that financial institutions should have an incentive to link with remittances since studies show that remittances has an important part in development. Remittances help distribute capital, thus increase the life situation for poor individuals and by financial intermediation the capital could be used more efficiently. If individuals who receive remittances bring them first to financial institutions instead of consuming it they will have a link to banks that can make them creditable in the future.
According to Gheeraert and Sukadi Mata (2008) remittances have two impacts, one on transaction cost and one on openness. They have found that remittances are most used to investments if transactions cost (access to local banks) are low. This might even induce individuals to deposit the remittance they receive and by that create funds that can be lent out to investments. Remittances are also more important for investment if openness to other financial institutions is restricted i.e. at what level banks are open internationally. If international banks have information about the domestic ones, risk premiums can be reduced. This information is more likely to be available if the financial sector is more developed and as a result the impact of remittances decreases. To summarize; there is a positive relationship between transaction costs and investments and there is a negative relationship between openness and investments when it comes to remittances (Gheeraert & Sukadi Mata, 2008)
Infrastructure is a complex field and can be defined both as material and non-material. This issue makes it hard to conclude what infrastructure is supposed to contain.
Garcia-Milà and McGuire (2008) as well as Andersson and Strömquist (1988) claim that a country’s infrastructure is essential for an economy and the better it is, the easier it will be for the economy’s actors, such as firms in the private sector, to interact with each other. It will also help to equalize the economy, by balancing prices in regions. The frequent view on today’s economic life is that production plays the central role for a firm, with a large capital share, while communication and transportation comes behind. However, communication and transportation are still the basic needs for a firm, and these factors make production possible (Andersson & Strömquist, 1988).
Bircenño-Garmendia, Estache and Shafik (2004) also point out the necessary support of infrastructure to a developing country. Infrastructure, together with financial and economic policies is accepted key elements for economic growth. The authors include roads, telecommunication and electricity in their definition of the subject, and claim that these are all rather poor developed in low-income countries. A common measurement of roads is the percentage of paved roads to total roads. Paved roads are a more efficient way to transport goods, and a network of paved roads instead of dirt roads is more reliable from a firm’s point of view. However, even though the World Bank has good statistics on the percentage of paved roads, the authors claim that this variable might be to some extent misleading. This is because paving roads is not the most important priority for a developing country and poor people might value fresh water and quality of treated sewage more (Bircenño-Garmendia, Estache and Shafik, 2004).
Compared to other sorts of capital, which are mostly reached on a local basis, communication systems can build networks and channel mobility. Andersson and Strömquist (1988) argue that the development of an infrastructure with knowledge and competence and a material infrastructure with networks of different kinds, are fundamental for the development of how fast and in what form the economy will grow. Transportation networks and road penetration is an important part and the use of roads is essential for trade. Government’s engagement (its public spending) is important for the expansion of road networks. A positive relationship between road development and population density is widespread, however not present everywhere (Andersson & Strömquist, 1988). For example, in north of Ghana the cocoa districts have a weak road network and Taaffe, Morrill and Gould (2006) suggest that this could be due to political reasons, such as conflicts between regions.
Material infrastructure can be roads and railways, building structures and water pipes while non-material infrastructure can be institutions, education, telecommunication and other connecting networks. The complexity of this subject makes it hard to handle and the importance of both material and non -material infrastructure can make it somewhat confusing. Non-material infrastructure is more important today than it was in earlier years, due to technology, such as telecommunication. Yang and Parent (2008) claim that the importance of these new connecting networks puts pressure on developing countries to increase and improve, not only the material infrastructure such as roads and railways, but also to invest in new technology, such as telephone lines and Internet. These new demands of technology investments are hard for poorer countries to meet, and the way to an industrial level becomes more distant.
Telecommunication is also a part of a country’s infrastructure and is important for a country’s development since it among other things gives access to information and Internet, and it makes it easier for business to have contact with a larger market. Bohman (2008) has studied telecommunications, in relation to income distribution in Brazil. The findings of the article are that an uneven distribution of income in poor regions can be beneficial since that implies that some people in the region afford to take part in services like telecommunication. If there is even distribution instead it implies that no-one in the region has the income necessary to afford those kind of services. The opposite relation holds for richer regions (Bohman, 2008).
Table of Contents
2.2 Firm start-up and survival
2.3 Labor Market Development
2.4 Credit Access for Poor in the Agricultural Sector
3.1 Microeconomic Theory of Firm start-up
3.2 Importance of local access to capital
4.1 Previous Hypotheses and Their Testing
4.2 A Revised Hypothesis
4.3 Data collection
5 Result and analysis
6 Conclusions and Suggestions for Further Studies
GET THE COMPLETE PROJECT
Microcredit and Firm Start-ups and Survivals in Developing Countries