IMPORTANCE OF ACCESS TO FINANCE IN REDUCING INCOME INEQUALITY AND POVERTY LEVEL

Transcription

1 International Review of Public Administration 2012, Vol. 17, No IMPORTANCE OF ACCESS TO FINANCE IN REDUCING INCOME INEQUALITY AND POVERTY LEVEL KWANGBIN BAE Seoul National University, South Korea DONGSOOK HAN University of California, Los Angeles, USA & HOSUNG SOHN University of California, Berkeley, USA This study investigates the relationship between access to finance, and poverty and income inequality. We first define access to finance and identify various measurements of access to finance. Next, we examine previous analyses on the impacts of access to finance. Finally, using statelevel panel data of the United States, fixed effect estimation is conducted to analyze the impact of access to finance on income inequality and poverty level. Our analysis is the first study to utilize state-level data on access to finance. The results show that access to finance has positive effects in reducing income inequality and the poverty ratio. Key Words: Access to Finance, Income Inequality, Poverty Level, Fixed Effect Model INTRODUCTION Nowadays, financial development 1 is regarded as an important factor that determines a nation s economic growth. This is because finance plays an important role in boosting a national economy by allocating resources efficiently in the process of accumulating and pooling capital (Rajan and Zingales, 1998; Beck, Levine, and Loayza, 2000;

2 56 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 Chakraborty and Ray, 2006; Deidda and Fattouh, 2006). With respect to its effect on income inequality, however, previous research is not unanimous. Recent studies argue that financial development not only provides an engine for economic growth, but also promotes income equality and the welfare of poor people (Clark, Xu, and Zou, 2006; Beck, Demirguc-Kunt, and Levine, 2004; Clark, Xu, and Zou, 2003; Beck, Demirguc- Kunt, and Levine, 2007; Honohan, 2004). On the other hand, though in the minority, some researchers maintain that financial development has a negative impact on income equality (Greenwood and Jovanovi, 1990; Halac and Schmukler, 2003) What is the reason for the existence of these two conflicting conclusions? In order to address this question, we must first understand how financial development promotes income equality. When a financial system is underdeveloped, people in lower income brackets are more likely to be located geographically far away from formal financial institutions. Additionally, they often cannot afford the minimum balance required to open a transaction account or a savings account. Furthermore, they are unable to use financial services such as deposits, payments, insurance, and credit because the overdraft penalties are too high (Claessens, 2006; Honohan, 2008). However, as a financial sector develops, these transaction costs and information costs will decrease. Accordingly, development of the financial sector would raise the possibility of access to finance for low-income groups (Burgess and Pande, 2005; Demirguc-Kunt and Levine, 2008). In order for financial development to be positively associated with increasing income equality and decreasing poverty, the degree of access to financial services must increase. Therefore, the important factor for reducing income inequality is not financial development itself. Rather, increasing access to finance is more critical to raising the income of the poor. Hence the focus of research should shift from the impact of financial development on income inequality to the effect of access to finance on income equality. Access to finance is characterized by an absence of barriers in the use of financial services. However, measures of financial development do not reflect this access dimension (Demirguc-Kunt and Levine, 2008). It is not necessarily true, moreover, that access to finance increases as a financial sector develops. Previous research investigating the relationship between financial development and income inequality has measured its development in terms of the depth of the financial system. The depth of a financial system is measured by private credit, 2 and it is generally limited to account for the measurement of access to finance (Demirguc-Kunt and Levine, 2008). Since previous research has failed to investigate the relationship between access to finance and income inequality, this study will focus on analyzing the effect of access to finance, not financial development itself. Furthermore, the effects of access to finance on poverty reduction will also be examined. To fulfill this goal, it is necessary for us to estimate the degree of access to finance. Recently, there have been attempts to define and measure various aspects of access to finance. 3 Moreover, some researchers have attempted to estimate the effect of access to finance

3 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 57 on income inequality and poverty reduction. The scope of their research, however, is cross-country data. In this study, state-level panel data will be used in the analysis. Using state-level data has methodological advantages over using cross-country data. First, compared to countries, states share more homogenous characteristics; therefore, estimating the exact impact of access to finance is more viable. Second, cross-country data on income inequality are not consistent enough to compare (Menard, 1986; Hoover, 1989). On the other hand, state-level data have been collected and calculated solely by the Census Bureau, which means that the data will be consistent and accurate. This study is organized as follows. First, we present some previous studies that have attempted to measure access to finance. Next, we will examine theoretical mechanisms that demonstrate how the restriction of access to finance will result in increased poverty and income inequality. Third, we will examine previous empirical analyses on the impact of access to finance on poverty and savings of the poor. Finally, using state-level panel data from the United States, we will empirically analyze the effect of access to finance on income inequality and the poverty level. DEFINITION OF ACCESS TO FINANCE Access to finance refers to the availability of reasonable quality of financial services with reasonable costs (Claessens, 2006). Accordingly, broad access to financial services implies that there is no price or no price constraint for using financial services. This concept is similar to the concept of financial inclusion, 4 in which every individual is able to use basic financial services. However, since the concept of access to finance covers various dimensions, it is quite hard to define its concept explicitly. Nowadays, several countries engage in surveying access to finance. However, because of the difficulty in defining access to finance, each country uses different kinds of representativeness or measurement in estimating access to finance. This inconsistency in defining access to finance makes it difficult to analyze cross-country comparisons. The study of access to finance is challenging also because access is difficult to observe (Demirguc-Kunt, Beck, and Honohan, 2008). As a result, most research focuses instead on use of financial services. Nonetheless, access is not always equal to use, so access and use of financial services should be classified as separate concepts. While use indicates consumption of financial services and hence is related to the demand side, access comprises both the demand and supply sides of financial services. That is, the concept of use is a subset of access. For example, if a person is voluntarily excluded from using financial services because he or she prefers nonfinancial means of transaction or because of cultural or religious reasons, this person is not using financial services, but has access to finance. However, a person can be involuntarily excluded from using financial services because of their unaffordability or because the person was subject to

4 58 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 discrimination. In such cases, we can say the person does not have access to finance. Thus, availability of services (access) is a necessary condition to use, not a sufficient condition. The difference between access and use is explicitly represented in Figure 1. People in B, C, E, and F have access to financial services. Contrarily, people in A, D, G, H, and I do not have access to financial services. Figure 1. Distinction between Access to Finance and Use Regardless of their distinct characteristics, it is difficult to separate these two concepts and only measure access. Recently, however, many researchers have recognized the importance of distinguishing access to finance from use of finance and have attempted to develop indicators for and to measure access to finance. Honohan (2008) is the representative of those who are engaged in estimating access to finance using proxy indicators, and he has estimated access percentage based data from 160 countries. The estimated access percentage is constructed as a function of the estimated number of bank accounts and average deposit size. When the estimated number of bank accounts was not available for a specific country, he used the microfinance account numbers instead. Moreover, when the average deposit size was unavailable, an estimate of average deposit size was generated as a function of GDP per capita. While Honohan (2008) used the use aspects of data when estimating the access percentage, Beck, Demirguc-Kunt, and Peria (2007) are different in the sense that they utilized both the availability and use aspects of data in their attempt at making new financial sector outreach. They derived an aggregate indicator by collecting data from 99 countries with respect to access to banks and use of

5 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 59 banking services that were obtained by surveying the bank regulators of each country. The survey consists of two large parts: the outreach of the financial sector in terms of the availability of a bank s physical outlets and the use of banking services. Another way to estimate access to finance is to estimate the barriers to financial services. Genesis (2005) investigated the cost of opening financial accounts in Brazil, India, Kenya, Malaysia, Mexico, Nigeria, and South Africa. However, this study is limited in that it was restricted to the investigation of deposit service affordability and in that he only surveyed seven countries. Demirguc-Kunt and Peria (2008) conducted a survey of up to five large banks per country in more than 80 countries. It was a first effort to document access barriers around the world. They created barrier indicators for three types of banking service: deposit, loan, and payment, and across three dimensions, physical access, affordability, and eligibility. Theoretical Research REVIEW OF PREVIOUS RESEARCH Typical theoretical studies aimed at analyzing the effect of access to finance on income inequality contend that financial market imperfections are the most crucial reason for the creation of inequalities in income and the persistence of poverty. These imperfect financial market environments play a central role in ensuring poor people remain in poverty for the following reasons. First, one study argues that financial market imperfections affect the level of education or the level of human capital of the poor, thereby promoting persistent poverty. According to Demirguc-Kunt and Levin (2008), under perfect financial market conditions, social efficiency can be attained in the sense that people with academic desires receive sufficient schooling regardless of parental wealth. This indicates that one s economic opportunity is totally dependent on one s ability. Contrarily, under imperfect financial market conditions, such as informational asymmetries, transaction costs, and contract enforcement (Beck et al., 2007), schooling is affected not only by academic desire, but also by parental wealth. More specifically, poor parents need to borrow money to afford schooling for their children, but imperfect financial markets create barriers to financing education. Accordingly, even though the poor may be academically capable, they have a hard time receiving schooling. On the other hand, comparably less capable children with more affluent parents tend to receive an excessive amount of education. Under imperfect financial market conditions, one s level of human capital and economic opportunities are determined not only by one s ability, but also by parental wealth. All in all, the authors contend that because of the barriers to financing an

6 60 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 education faced by poor people, income inequality and poverty levels are worsening. There are a few studies that underscore the fact that poverty persists because the financial market s imperfections determine who can become an entrepreneur, thereby affecting a person s occupation. According to the model developed by Banerjee and Newman (1993), and Agion and Bolton (1997), when a financial market functions perfectly, people with entrepreneurial abilities can become entrepreneurs whether they are rich or poor because everybody has equal access to the required funding. In an imperfect financial market, however, the poor cannot raise the external funds that are essential for starting a new business because most lenders require higher amounts of collateral. Therefore, even if a poor person has a good idea for a business, he or she cannot become an entrepreneur and must remain a wage laborer. To put it differently, an individual s occupational choice is solely dependent on his or her initial endowments (dynastic assets), in an imperfect financial market Empirical Studies A number of empirical studies investigate the impact of access to finance on the welfare of the poor. In general, these studies show that increases in access to finance positively affect the savings and income of poor people and thereby reduce the poverty level. The following research analyzes the effect of access to finance on the savings of poor people. Aportela (1999) pointed out that one reason people with low levels of income cannot save is because there is a barrier to using formal instruments. By analyzing Mexican households income and expenditures in a survey that was conducted in 1992 and 1994, he found out that an increase in Mexican saving institutions raised the average savings rate of the affected household by 3% to 5%, and the savings rate of lowincome households increased by about 8%. This study shows that if poor people are provided with relevant financial instruments, they can become regular savers. Contrary to the two aforementioned studies, Cigno and Rosati (1992) contend that as access to finance increases, the savings rate of the poor will decrease along with their fertility rate. Using savings rate data from 1960 to 1984 along with fertility rate data from 1953 to 1984, they observed that as capital market accessibility (estimated by the ratio of currency held by the non-banking public to M2) increased, the savings and fertility rates of an individual decreased. This study, however, was conducted during a period when the concept of access to finance and the methods for measuring it had not been developed, so there is limited use in interpreting its results. There has been some previous research on the effect of access to finance on the poverty level. Burgess and Pande (2005) analyzed the effect of an increase in the number of bank branches on the poverty level. They evaluated the effect of the enforcement of the bank branch expansion program that took place in India from 1969 to 1990 on the rural poverty level. 5 In order to estimate the rural poverty level, two dependent variables

7 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 61 Table 1. Summary of Previous Research Author Content Result Theoretical Analysis Demirguc-Kunt and Levin (2008) Banerjee and Newman (1993); Agion and Bolton (1997) In an imperfect financial market, because of the barriers to financing education that are faced by poor people, income inequality and the poverty level are getting worse. An individual s occupational choice is solely dependent on their initial endowments under imperfect financial markets, and the poor are excluded from investment opportunities that can bring a higher rate of return, which can force the poor to remain in poverty. Empirical Studies Aporetela (1999) Increases in access to finance positively affect the savings and income of poor people. Positive Relationship Burgess and Pande (2005) Khandker (2005) Beck, Levin, and Lekov (2007a, 2007b) Increases in the number of bank branches increase poor people s savings and mobilization. Microfinance program reduced the poverty level and raised the income of participants. Increases in the number of bank branches raised the efficiency of banks, and this in turn boosted the per capita income growth rate of each state. Positive Relationship Positive Relationship Positive Relationship Cigno and Rosati (1992) Increases in access to finance decreased the savings and fertility rate of the poor. Negative Relationship Note: Relationship in the Result category implies the relationship between access to finance and income of the poor were used: a headcount ratio determined by the ratio of population under the India poverty line and rural agricultural wages. The regression analysis revealed that as a result of bank branch expansions, poor people s savings mobilization and credit provision increased, implying that the poor accumulated capital. Additionally, the level of loans increased, enabling long-term investment. This in turn diminished the headcount ratio by 14% to 17% and raised rural agricultural wages, which decreased the rural poverty level. By analyzing the microfinance program that took place in Bangladesh in 1991 and 1999, Khandker (2005) analyzed the impact of access to finance on the poverty level. He found that in 1991, microfinancing reduced the poverty headcount by 5% by raising participants consumption levels, whereas in 1999, it was only reduced by 2%. Not only

8 62 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 did the program raise the income of participants, but it also raised the local income, and as a result it reduced the average village poverty level by 1%. Namely, the microfinance program had a spillover effect that was desirable for the local economy. This indicates that not only is an increase in access to finance desirable for the poor, but it is also favorable to economic growth. Beck, Levin, and Lekov (2007a, 2007b) conducted regression analysis to determine whether the increase in the number of bank branches in the United States had any impact on the income of the poor. The deregulation policy (appeasement of the restrictions on bank branch expansions in the United States) that was instituted in each U.S. state from 1970 to the middle of the 1990s was one of the reform policies aimed at promoting bank services by strengthening the competition among bank industries. According to Beck et al. s study, the increase in the number of bank branches raised the efficiency of banks, and this in turn boosted the per capita income growth rate of each state. The results of the existing research are provided in Table 1. DATA AND EMPIRICAL STRATEGY In this section, measures of financial access, income inequality, poverty, and econometric methods for ascertaining the relationship between these elements will be presented. The units of analysis are the fifty states of the United States of America from 2000 to Data As previously mentioned, access to finance can be measured using various methods and bases. In this study, I will employ three variables that are used by Beck, Demirguc- Kunt, and Peria (2007). First, in order to reflect the access to financial institutions various aspects, the following two indices have been used: the number of institutions per 10,000 people (demographic institution penetration) and the number of institutions per 100 square miles (geographic institution penetration). Next, to account for the use of financial service aspects, the average size of deposits to GDP per capita (deposit-income ratio) has been used. The number of institutions per 10,000 people can be interpreted as the number of people per financial institution. Therefore, higher numbers in this index imply easier access for greater demographic penetration, which would indicate fewer potential clients per institution. A greater number of institutions per 100 square miles can likewise be interpreted as increased access to finance. This is because greater geographic penetration would indicate less distance from and easier geographic access to institutions. The ratio of average size of deposit to GDP per capita indicates how much each individual possesses as a deposited balance. According to Beck, Demirguc-Kunt, and

9 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 63 Peria (2007), if this number is higher, poor people and smaller enterprises are less able to make use of financial services. 6 Accordingly, higher numbers in this index imply less access to finance. The three aforementioned indices are all easy to understand and interpret, and hence are widely used as representative indicators for estimating access to finance. However, these indicators are not without shortcomings. Above all, the ratios of the number of institutions to area and population assume a uniform distribution of bank outlets within a state s geographical area and across its population. This is limited in the sense that most states financial institutions are concentrated in the urban centers of those states. As to average size of deposit, the deposit is only one service among numerous financial services that individuals receive from financial institutions, and thus might not be representative of the use of finance. Nevertheless, as proven by previous research, these indices are highly correlated with the actual percentage of households and firms that use banking services, and it is reasonable to assume that there are few problems caused by using these indices to account for the degree of access to finance. The data on number of institutions and the deposit size used in this analysis were obtained from the Bank Data and Statistics provided by the Federal Deposit Insurance Corporation (FDIC). For the purposes of this paper, institutions indicate the financial institutions headquartered in a state that have been granted legal authorization to conduct business by the federal or state government. These include commercial banks, 7 savings institutions, 8 and U.S. branches of foreign banks insured by the FDIC. 9 To assess the impact of financial access on the poor, this study will analyze the impact of financial access on income inequality and poverty. First, as a dependent variable, the Gini coefficient is used to account for income inequality. Furthermore, to investigate the robustness of the analysis, sensitivity analysis will be conducted using other indicators of income inequality. Because of the data availability, this study will use 1) the Theil index and 2) the Atkinson index. In order to find out whether increased access to finance alleviates poverty, 3) the ratio of households below the poverty level (poverty ratio) will be used as another dependent variable. Data on these four variables were acquired from the American Community Survey (ACS, ) conducted by the U.S. Census Bureau. In Figure 2, time-series plots of income inequality indicators and the poverty level have been depicted together with the access to finance variables. As can be seen from the figure, while income inequality and the poverty level tend to worsen over the period, access to finance decreases, implying that there are negative associations between these two sets of indicators. 10

10 64 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 Figure 2. Time-Series Plot of Key Variables Estimation Methods In this paper, state-level data will be used to test the relationship between access to finance and income inequality and poverty. Furthermore, since data are available over time, multiple years of state-level data will be employed for empirical analysis. When the data have space as well as time dimensions, panel data regression models should be conducted, because a combination of time-series and cross-section data provides more informative data, more variability, less collinearity among variables, more degrees of freedom, and more efficiency. Moreover, panel data are better suited for studying the dynamics of change. There are other advantages to using panel data, so if appropriate data are available, use of panel regression produces numerous meaningful results. While there are various methods of panel regression, researchers, in general, choose between the fixed effect model (FEM) and the random effect model (REM). Although there are many factors to consider in choosing between these two models, REM should only be used whenever one is confident that its composite error is not correlated with the explanatory variables. One way to test this is to use the Hausman test, which tests whether the REM estimate is significantly different from the FEM (Kennedy, 2008). The Hausman test is calculated as an F-test for testing coefficients estimated by both models against zero. In these datasets, results of the Hausman test suggest that we should use FEM, since the 2 statistic is Hence, the model used to analyze the impact of access to finance on income inequality and poverty is n y it = 1 it + kz kit + i + it k=1

11 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 65 where i stands for the i-th cross-sectional unit, t for the t-th time period, x refers to the proxy for financial access variables, Z is the vector for other independent variables that are used for controlled variables, y will be either income inequality measures or a poverty variable, and and are an unobserved fixed effect and an error term, respectively. Table 2. Summary of Variables Variable ID Definition Source Scale Demographic institution PINST The number of financial institutions per FDIC Unit penetration 100,000 people Geographic institution MINST The number of financial institutions per FDIC Unit penetration 100 square miles Deposit-income ratio DEPOS Average size of deposits to GDP per capita FDIC % Gini coefficient GINI One measure of income equality, ranging ACS % from 0 to 1 Theil s index THEIL A measure of inequality where indicates ACS % perfect inequality and 0 indicates perfect equality Atkinson index ATKIN The inequality index having a potential range ACS % from 0 to 1. The more equal the income distribution, the lower the value. Ratio of households POV Ratio of households to total households whose ACS % below poverty level income in the 12 past months is below poverty level Natural logarithm of MED Median family income of each state ACS $ median family income (inflation adjusted) Educational attainment EDU Ratio of population 18 years and over who ACS % received associate s, bachelor s, graduate, or professional degree Female-labor force FELP The percent of females in the labor force ACS % participation (population 16 years and over) Percentage of female- HEAD Ratio of female-headed families (no husband ACS % headed family present) to total number of families Percentage of population AGE Ratio of population over age 65 to total ACS % over age 65 population Manufacturing MANU Ratio of the employed civilian population 16 ACS % employment years and over in the manufacturing industry Property income ratio PROP Ratio of property income to total income ACS %

12 66 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 Table 3. Estimate of Income Inequality State Gini Coeffi. Theil Index Atkinson Atkinson Atkinson Poverty ( =0.25) ( =0.50) ( =0.75) Ratio Alabama (5) (7) (7) (6) (4) (6) Alaska (50) (50) (50) (50) (50) (48) Arizona (22) (19) (19) (19) (19) (18) Arkansas (23) (22) (21) (21) (23) (7) California (6) (6) (8) (9) (9) (27) Colorado (28) (29) (27) (27) (29) (39) Connecticut (3) (2) (2) (3) (5) (49) Delaware (9) (41) (41) (40) (39) (44) Florida (8) (3) (4) (7) (10) (24) Georgia (15) (15) (15) (15) (13) (15) Hawaii (38) (37) (35) (32) (28) (38) Idaho (44) (44) (12) (45) (43) (20) Illinois (13) (8) (10) (10) (11) (30) Indiana (41) (42) (42) (42) (41) (33) Iowa (46) (47) (47) (47) (47) (34) Kansas (33) (31) (31) (33) (34) (28) Kentucky (14) (17) (16) (16) (15) (4) Louisiana (2) (4) (3) (2) (2) (2) Maine (39) (39) (38) (38) (38) (22) Maryland (34) (35) (33) (34) (33) (47) Massachusetts (10) (9) (9) (8) (7) (37) Michigan (29) (33) (32) (30) (30) (26) Minnesota (42) (43) (43) (43) (44) (45) Mississippi (7) (10) (6) (5) (6) (1) Missouri (27) (28) (29) (31) (32) (21) Montana (36) (36) (37) (37) (37) (13) Nebraska (40) (38) (39) (39) (40) (32) Nevada (32) (20) (22) (25) (26) (40) New Hampshire (48) (48) (48) (48) (48) (50) New Jersey (17) (16) (17) (18) (18) (46) New Mexico (11) (13) (12) (12) (17) (5)

13 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 67 New York (1) (1) (1) (1) (1) (14) North Carolina (18) (14) (14) (14) (16) (12) North Dakota (37) (34) (34) (36) (36) (17) Ohio (30) (32) (28) (28) (27) (23) Oklahoma (19) (18) (18) (17) (12) (8) Oregon (26) (23) (24) (26) (25) (16) Pennsylvania (21) (21) (20) (20) (21) (29) Rhode Island (25) (25) (25) (23) (22) (25) South Carolina (16) (11) (11) (11) (8) (10) South Dakota (35) (30) (34) (35) (35) (19) Tennessee (12) (12) (13) (13) (14) (11) Texas (4) (5) (5) (4) (3) (9) Utah (49) (49) (49) (49) (49) (43) Vermont (43) (45) (44) (44) (46) (35) Virginia (24) (24) (23) (22) (24) (42) Washington (31) (27) (30) (29) (31) (31) West Virginia (20) (26) (26) (24) (20) (3) Wisconsin (47) (46) (46) (46) (45) (41) Wyoming (45) (40) (40) (41) (42) (36) Note: The estimates are the average of Numbers in parentheses are the ranking of each state. In accordance with interstate income inequality literature (Nielsen and Alderson, 1997; Bishop, Formby, and Thistle, 1992; Braun, 1988), this study controls for median family income, educational attainment, the percentage of the population over age 65, the percentage of females in the labor force, the percentage of female-headed households, the percentage of labor force in manufacturing, and the percentage of property income. All data were obtained from the ACS. As with income inequality, the time ranges from 2000 to The definition, sources, scale, and identification of all the aforementioned variables are shown in Table 2. Table 3 shows the income inequality indices and poverty ratio for each state, and the estimate shown in the table is the average of the estimate of 2000 to 2007.

14 68 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 RESULTS The regression results are represented in Table 4, Table 5, and Table 6. Table 4 shows the FEM estimates in which Gini and Theil indices are used as dependent variables. For Table 5, the Atkinson index for three values of c is used as the dependent variable. In Table 6 the poverty ratio is the dependent variable for FEM estimates. For each dependent variable, three kinds of access variable have been used to measure the impact of access to finance on income inequality (Model 1 to Model 3). Definitions of all the variables used in the table are provided in Table 2. Impact on Income Inequality According to Table 4, access to finance as it relates to demographic and geographic financial institution penetration does not have an impact on Gini coefficients. Neither of these estimates was significant at the 10% level. On the other hand, the deposit-toincome ratio indicates a significant positive correlation between access to finance and Gini coefficients. As was previously mentioned, Beck, Demirguc-Kunt, and Peria (2007) assert that if the deposit-to-income ratio is high, poor people and small enterprises are less likely to make use of financial services. Hence, the results in Table 4 imply that the degree of access to finance decreases as the deposit-to-income ratio rises, while, at the same time, the Gini coefficient increases, which in turn indicates that income inequality has worsened. Three variables turned out to be highly significant with respect to their ability to affect the Gini numbers: (1) the percentage of the population over age 65, (2) the level of manufacturing employment, and (3) the percentage of families headed by women. Previous studies show that people receive relatively higher wages when working in the manufacturing sector. Moreover, it has been discovered that wage distribution is more equal in this sector than in other sectors (Bluestone, 1990). Therefore, in this last regard, our results are equivalent to those seen in the previous research on this subject, in the sense that the variable we used is negatively correlated to income inequality. Estimations of FEM using the Theil index as an explained variable are also shown in Table 4. Contrary to the FEM of the Gini coefficient, the deposit-to-income ratio turned out to have an insignificant effect on income inequality. However, if the degree of access to finance is defined by the accompanying degree of geographic financial institution penetration, the effect of that access is thereby derived to have a negative impact on income inequality. Although the significance level for this variable is slightly less than 10%, it still shows that access to finance promotes income equality. Similarly, the three variables that were significant in our Gini models all turned out to be significant in our Theil models. Moreover, all of their effects are equivalent to those found in the Gini models except that the significance level for the variable is, specifically, different.

15 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 69 Table 4. FEM (Dependent Variable: Gini and Theil Indices) Gini Coefficient Theil Index Independent Variable Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Demographic institution penetration (0.021) (0.032) Geographic institution * penetration (0.010) (0.015) Deposit-income ratio 0.022*** (0.003) (0.005) Percentage of population 0.889*** 0.855*** 0.899*** 1.204*** 1.066** 1.161*** over age 65 (0.293) (0.295) (0.272) (0.438) (0.439) (0.438) Manufacturing employment 0.202*** 0.217*** 0.210*** 0.194* 0.226** 0.219** (0.070) (0.069) (0.064) (0.104) (0.102) (0.103) Percentage of female 0.284*** 0.268*** 0.262*** 0.368*** 0.328** 0.342** headed family (0.090) (0.090) (0.083) (0.135) (0.134) (0.134) Natural log of median family income (0.018) (0.017) (0.016) (0.026) (0.025) (0.026) Female-labor force participation (0.103) (0.103) (0.096) (0.153) (0.153) (0.154) Educational attainment (0.311) (0.311) (0.289) (0.381) (0.343) (0.305) Educational attainment squared (0.490) (0.491) (0.456) (0.732) (0.731) (0.735) Property income ratio (0.062) (0.059) (0.054) (0.092) (0.088) (0.088) Intercept (0.178) (0.164) (0.155) (0.266) (0.245) (0.250) Number of observations Note: Number in parentheses is the standard error. ***p < 0.01, **p < 0.05, * p < 0.10 Table 5 shows the FEM estimates, where the Atkinson index for three 11 is used as a dependent variable. As with the Theil index, the access to finance when defined as the degree of geographic financial institution penetration came out to be negatively correlated with income inequality. That is, the greater the number of financial institutions in a certain region, the more equal the distribution of income. The negative impact of this

16 70 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 Table 5. FEM (Dependent Variable: Atkinson Index) Independent Variable Atkinson ( =0.25) Atkinson( =0.50) Atkinson( =0.75) Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Model 1 Model 2 Model 3 Demographic institution penetration (0.006) (0.010) (0.015) Geographic institution penetration 0.007** 0.012** 0.015** (0.003) (0.005) (0.007) Deposit-income ratio (0.000) (0.002) (0.002) Percentage of population over age *** 0.259*** 0.258*** 0.492*** 0.438*** 0.481*** 0.664*** 0.594*** 0.649*** (0.086) (0.085) (0.086) (0.144) (0.143) (0.143) (0.201) (0.200) (0.200) Manufacturing employment 0.062*** 0.068*** 0.066*** 0.120*** 0.129*** 0.126*** 0.183*** 0.195*** 0.191*** (0.020) (0.020) (0.020) (0.034) (0.033) (0.034) (0.048) (0.047) (0.047) Percentage of female-headed family 0.076*** 0.068*** 0.072*** 0.137*** 0.124*** 0.131*** 0.194*** 0.178*** 0.187*** (0.026) (0.026) (0.026) (0.044) (0.044) (0.044) (0.062) (0.061) (0.062) Natural log of median family income (0.005) (0.005) (0.005) (0.009) (0.008) (0.008) (0.012) (0.012) (0.012) Female-labor force participation (0.030) (0.030) (0.030) (0.050) (0.050) (0.051) (0.070) (0.070) (0.071) Educational attainment (0.091) (0.090) (0.091) (0.152) (0.151) (0.153) (0.213) (0.211) (0.213) Educational attainment squared (0.143) (0.142) (0.143) (0.240) (0.238) (0.241) (0.336) (0.333) (0.336) Property income ratio (0.018) (0.017) (0.017) (0.030) (0.028) (0.029) (0.042) (0.040) (0.040) Intercept (0.052) (0.048) (0.049) (0.087) (0.080) (0.082) (0.122) (0.117) (0.114) Number of observations

17 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 71 variable on the Atkinson index turned out to be consistent for all the models used. One thing to note here is that when society places greater importance on income equality, geographic financial institution penetration has a greater impact on reducing income inequality. The estimate of Atkinson index for = 0.75 is two times larger than the estimate of Atkinson index for = This can be interpreted to mean that if, in a certain region, people attach more weight to income transfers for those in the low-income brackets, the probability that poor people there will use financial institutions will increase compared to those instances when the poor are in a state or region where people do not care about income inequality. This is why the effect is greater if a higher? is used for the estimation. Apropos of this, the geographical penetration variable turned out to be significant at the 5% level in all models. Impact on Poverty Ratio In Table 6, the ratio of households below the poverty level was used as a dependent variable when estimating the FEM. While access to finance as it relates to the degree of demographic financial institution penetration neither increased nor decreased income inequality, we discovered that it effectively reduced the poverty ratio, thereby conforming to the assertion, made in previous research, that there is a negative correlation between access to finance and the poverty level. Additionally, a positive correlation between the deposit-to-income ratio and the poverty ratio has also been deduced, a finding that also coincides with the hypothesis verified in previous research. Both of these estimates coincide with the hypothesis that access to finance reduces the poverty level. Again, both the significance level and the sign of coefficient of the three variables (1) the percentage of the population over 65, (2) the level of manufacturing employment, and (3) the percentage of female-headed families are all the same. That is, as the percentage of elderly and female-headed households rises, so does the level of poverty. In contrast, as the ratio of population employed in manufacturing increases, the level of poverty decreases. There are two additional variables that proved to be significant in Table 6: the ratio of property income to total income, and median family income. As was previously mentioned, Braun (1988) determined that the property income ratio is positively correlated with various measures of income inequality. While this may normally be the case with income inequality, the result of our study shows that the property income ratio is negatively correlated with the poverty ratio. This variable is highly significant and its effect is to imply that as the property income ratio rises, the poverty ratio falls. Such a result is reasonable in that, logically, the property income ratio is likely to have different effects on income inequality, on the one hand, and the poverty level, on the other. Since property income refers to any income deriving from interest, rent, or dividends, it is

18 72 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 likely that an increase in this type of income would overwhelmingly benefit the rich, and, thus, worsen the income gap. Contrarily, an increase in property income would not have an impact on the poverty level because poor people seldom if ever have this kind of income, and hence the poverty ratio would not be affected in any noticeable way by changes in the property income ratio. On the other hand, tax revenues would be likely to rise if there were an increase in the property income ratio. As a consequence, government would be in a position to possibly use the increased tax revenue to help the poor. Hence, as the results of this analysis suggest, an increase in property income might imaginably be of some benefit to the poor, depending on the political complexion of the government at any given time and its view on income redistribution. Table 6. FEM (Dependent Variable: Poverty Ratio) Independent Variable Model 1 Model 2 Model 3 Demographic institution penetration 0.032** (0.012) Geographic institution penetration (0.006) Deposit-income ratio 0.005*** (0.002) Percentage of population over age *** 0.502*** 0.535*** (0.173) (0.175) (0.172) Manufacturing employment 0.193*** 0.178*** 0.175*** (0.041) (0.041) (0.040) Percentage of female-headed family 0.233*** 0.246*** 0.248*** (0.053) (0.053) (0.053) Natural log median family income 0.054*** 0.046*** 0.051*** (0.010) (0.010) (0.010) Female-labor force participation rate (0.103) (0.103) (0.096) Educational attainment (0.183) (0.184) (0.183) Educational attainment squared (0.289) (0.291) (0.289) Property income ratio 0.123*** 0.149*** 0.156*** (0.036) (0.035) (0.034) Intercept 0.646*** 0.541*** 0.589*** (0.105) (0.097) (0.098) Number of observations

19 April 2012 Kwangbin Bae, Dongsook Han, & Hosung Sohn 73 CONCLUSION In this study, an analysis of the impact of the access to finance on income inequality and the poverty ratio was performed by using data gathered at the state level (rather than nationwide) from 2000 to By utilizing various measures of income inequality and the poverty ratio, a panel regression using a fixed-effect model was conducted to test the effect of access to finance. Prior to performing the analysis, a thorough investigation of the definition of access to finance was conducted along with an inquiry into the previous research on the subject. While there are various ways to account for the apparent degree of access to finance, because of the issue of data availability, this study used three distinct variables: (1) the number of financial institutions per 100,000 people, (2) the number of such institutions per 100 square miles, and (3) the ratio of average deposit size to GDP per capita. For income inequality, the Gini coefficient, the Theil index, and the Atkinson index were used to reflect the sensitivity analysis. Our FEM summoned to mind several interesting implications. First, as in the previous research on this subject, the regression analysis showed that, in general, access to finance apparently has a positive impact by virtue of the fact that it reduces both income inequality and the poverty level. While not all three proxies were significant with respect to all the models used, at least one access indicator was always significant in each model, implying that access to finance always has some effect on income inequality and the poverty level. Moreover, the ratio of elderly and female-headed households to the total number of households as well the proportion of the population employed in the manufacturing sector proved to be at once strongly and consistently correlated with income inequality and the poverty ratio. One interesting thing to note is that in this study at least, both counter-intuitively and contrary to previous analyses, educational attainment had no effect on promoting income equality and on reducing the poverty level. This might be explained by the fact that our model might possess a certain kind of testing bias, such as multicollinearity or selection bias. Thus, further analysis should be pursued in order to estimate the impact of educational attainment. All in all, the policy implication of this study seems straightforward. Recent studies contend that when access to finance is not secured for the poor, a mere increase in financial development is limited in its ability to promote income equality and to reduce poverty. While previous research on this issue employed cross-national analysis in order to investigate the effect of access to finance, this study is the first attempt to investigate the impact of such access that is specifically based on state-level data. As the outcome of our analysis seems to show, the results of a study that chooses to examine state-level data are no different than those of studies conducted cross-nationally with respect to the impact of access to finance. While additional studies should be conducted to ascertain the effects of financial access, this study shows that in order to promote income equality and reduce poverty, effective policies should be created that are targeted at improving access

20 74 Importance of Access to Finance in Reducing Income Inequality and Poverty Level Vol. 17, No. 1 to finance for the poor. NOTES 1. The concept of financial development indicates the degree to which the financial system ameliorates information and transaction costs and facilitates the mobilization and efficient allocation of capital (Beck, Demirguc-Kunt, & Levine, 2004). 2. Private credit equals the logarithm of claims of financial institutions on the private sector as a share of GDP averaged over a sample period (Beck, Demirguc-Kunt, & Levine, 2007). 3. Section 1 in this paper will deal with this matter. 4. A financial system where many people can use financial services. 5. In 1969, right after bank nationalization, the Indian government enforced a program targeted at increasing the number of bank branches. This program was aimed at promoting the welfare of poor people living in rural areas by raising their access to formal credit and providing opportunities to save their money. 6. Affordability of deposit services is characterized by the minimum balance required to open checking accounts plus the fees to maintain the accounts. There is substantial variation across countries in the ratio of the minimum balance needed to open a checking account to GDP per capita. In Cameroon and Nigeria, the minimum balance to open a checking account exceeds 100 percent of per capita income, and in Ethiopia, Nepal, Sierra Leone, and Uganda, it is more than 50 percent, but in 18 countries, less than half of them developed, there is no minimum balance. 7. These institutions are regulated by one of the three Federal commercial bank regulators (FDIC, Federal Reserve Board, or the Office of the Comptroller of the Currency). They submit financial reports to the Federal Reserve (state member banks) or the FDIC (state nonmember banks and national banks). 8. This refers to savings institutions that operate under state or federal banking codes applicable to thrift institutions. They are regulated by and submit financial reports to one of two Federal regulators (FDIC or the Office of Thrift Supervision). 9. Branches of foreign banks include insured branches in the U.S. established by banks chartered and headquartered in foreign countries. These institutions are regulated by one of the three federal commercial bank regulators and submit financial data to the Federal Reserve. 10. Note that degree of access to finance is aggravating if the ratio of average deposit size to GDP per capita (third graph located at the below) is increasing. 11. The parameter reflects the degree of society s preference for equality, which ranges from zero to infinity. When > 0, it means that society prefers equality (or an aversion to inequality). As increases, society attaches more weight to income transfers at the lower portion of the income distribution and less weight to transfers at the top portion. Typically 0.5 or 2 is used for the values of.