Regarding the indicators which measure financial inclusion and its determinants, Table 2 provides descriptive statistics. We found that, overall, the account ownership rate of socially excluded groups such as women and the lower-income class is lower than average.
Table 2
Variable | Mean | S.D. | Min | Max | Count |
Account for all | 59.69 | 26.69 | 8.57 | 99.75 | 147 |
Account(%age 60+) | 57.85 | 28.87 | 7.40 | 100.00 | 147 |
Account(%age 35–59) | 50.01 | 26.30 | 7.00 | 100.00 | 147 |
Account for Female | 56.23 | 28.38 | 4.71 | 100.00 | 147 |
Account for Male | 63.10 | 25.38 | 12.50 | 99.85 | 147 |
Account for poorest 40% | 51.62 | 29.38 | 3.85 | 99.79 | 147 |
Account for richest 60% | 65.05 | 25.29 | 11.71 | 100.00 | 147 |
ATM | 49.94 | 47.46 | 0.37 | 272.82 | 147 |
Bank | 15.68 | 13.63 | 0.45 | 71.45 | 147 |
log(GDP) | 8.63 | 1.43 | 5.88 | 11.59 | 147 |
log(Internet) | 5.74 | 3.10 | 0.00 | 11.16 | 147 |
Rural | 39.96 | 22.21 | 0.00 | 83.65 | 147 |
Lower Secondary | 0.89 | 0.31 | 0.00 | 1.00 | 147 |
Upper Secondary | 0.54 | 0.50 | 0.00 | 1.00 | 147 |
Although some control variables had high correlations as shown in Table 3, multicollinearity was not a concern in our model. To test the existence of multicollinearity, we calculated the Variance Inflation Factors of the standard OLS model with all variables inserted. The individual values ranged from 1.63 to 6.99 and the mean value was 3.5; with no values above the rule-of-thumb cutoff of 10 (Neter et al. 1989) and very few values above 5, there was no multicollinearity among our variables.
Table 4 indicates the analysis results of the IV model showing the effect of compulsory upper-secondary education on financial inclusion. Column (1) shows the OLS results with only the control variable. The numbers of ATMs (b = 0.074, p = 0.049), banks (b = 0.146, p = 0.161), and Internet servers (b = 0.0002, p = 0.100) per one million people, which were used to check the accessibility of financial products and services, were positively related to financial inclusion. The numbers of ATMs and Internet servers were significantly positive, but the number of banks was insignificant. Like the findings in many previous studies (2), the coefficient of GDP is b = 12.735 (p = 0.0001) and is considerably positively related to the level of financial inclusion by country. Compulsory upper-secondary education has a positive correlation (b = 7.284, p = 0.039) with financial inclusion. For example, the rate of having an account is about seven percentage points higher in countries where upper-secondary education is compulsory.
Next, we analyzed the effect of compulsory upper-secondary education on financial inclusion by using instrument variables. Before the analysis, the following useful diagnostic tests of instrument validity were performed as stated in the lower part of (3) in Table 4: F-statistics, Wu-Hausman test, and Sargan test. In the F-statistics, our F-value was 23.425, with a significant p-value, indicating that our instruments were strong. The Wu-Hausman test showed that our OLS regression was not consistent, suggesting endogeneity was present. The Sargan test was not significant (p > 0.20), which indicated that our instruments were jointly uncorrelated with error terms, which showed their validity.
In Table 4, column (2) presents the result of the first stage model using the validated instruments. The p-values for Expected Years and Rights were low, indicating strong correlations between these instruments and the endogenous variable Upper Secondary after controlling for other variables. Column (3) is the result of the IV model showing the influence of upper-secondary education on financial inclusion. The coefficient of Upper Secondary is 26.411 (p = 0.0007), indicating a strongly positive effect, which supports our hypothesis. For example, the rate of owning an account is almost 26 percentage points higher for a person completing upper-secondary education.
Table 3
Correlation matrix between control variables.
Variable | (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | (10) | (11) | (12) | (13) | (14) |
(1) Account for all | 1 | | | | | | | | | | | | | |
(2) Account(%age 60+) | 0.921 | 1 | | | | | | | | | | | | |
(3) Account(%age 35–59) | 0.940 | 0.830 | 1 | | | | | | | | | | | |
(4) Account for Female | 0.991 | 0.914 | 0.934 | 1 | | | | | | | | | | |
(5) Account for Male | 0.990 | 0.909 | 0.930 | 0.964 | 1 | | | | | | | | | |
(6) Account for poorest 40% | 0.990 | 0.924 | 0.942 | 0.983 | 0.977 | 1 | | | | | | | | |
(7) Account for richest 60% | 0.994 | 0.906 | 0.925 | 0.983 | 0.986 | 0.967 | 1 | | | | | | | |
(8) ATM | 0.693 | 0.671 | 0.623 | 0.702 | 0.671 | 0.683 | 0.691 | 1 | | | | | | |
(9) Bank | 0.536 | 0.538 | 0.420 | 0.550 | 0.515 | 0.526 | 0.536 | 0.529 | 1 | | | | | |
(10) log(GDP) | 0.832 | 0.822 | 0.751 | 0.826 | 0.818 | 0.820 | 0.829 | 0.716 | 0.535 | 1 | | | | |
(11) log(Internet) | 0.572 | 0.535 | 0.582 | 0.587 | 0.549 | 0.590 | 0.550 | 0.421 | 0.362 | 0.599 | 1 | | | |
(12) Rural | -0.604 | -0.619 | -0.526 | -0.580 | -0.614 | -0.582 | -0.613 | -0.540 | -0.411 | -0.804 | -0.640 | 1 | | |
(13) Lower Secondary | 0.404 | 0.412 | 0.333 | 0.400 | 0.402 | 0.368 | 0.427 | 0.323 | 0.305 | 0.305 | 0.524 | -0.377 | 1 | |
(14) Upper Secondary | 0.673 | 0.656 | 0.571 | 0.674 | 0.658 | 0.652 | 0.678 | 0.640 | 0.469 | 0.469 | 0.728 | -0.642 | 0.333 | 1 |
Table 4
OLS and IV estimations for account ownership.
| (1) OLS | (2) First stage | (3) IV model |
Constant | -69.339*** (17.554) | -0.866** (0.396) | -64.102*** (19.433) |
ATM | 0.074** (0.037) | 0.002** (0.001) | 0.022 (0.045) |
Bank | 0.146 (0.104) | 0.003 (0.002) | 0.095 (0.116) |
GDP | 12.725*** (1.883) | 0.001 (0.044) | 10.881*** (2.171) |
Internet | 0.0002* (0.0001) | 0.0000 (0.0000) | 0.0002 (0.0001) |
Rural | 0.205** (0.091) | -0.003 (0.002) | 0.305*** (0.106) |
Upper Secondary | 7.284** (3.488) | | 26.411*** (7.657) |
Expected Years | | 0.105*** (0.015) | |
Rights | | -0.051** (0.022) | |
Weak instruments | | | 23.425*** |
Wu-Hausman | | | 10.850** |
Sargan test | | | 1.613, p = 0.2041 |
Adj. R-squared | 0.729 | 0.658 | 0.670 |
Observations | 147 | 147 | 147 |
Notes
This table reports the results of OLS and IV estimations for account ownership with instruments with Expected Years and Rights. Expected Years refers to the number of years of schooling a child of school admission age can expect to receive if prevailing patterns of average age-specific enrolment rates persist throughout the child’s life and Rights denotes the human rights index and the degree to which governments protect and respect human rights. ***, **, * denote significance at the 1%, 5%, and 10% level, respectively. Robust standard errors are in parentheses in all the models.
Next, we constructed the OLS and IV models with different individual characteristics. The results are shown in Table 5. The findings show that account ownership is positively related to compulsory upper-secondary education based on individual characteristics, except in populations aged 60 years and older. Column (1) shows the OLS regression results for the correlations of financial inclusion of elderly people. In the results, compulsory upper-secondary education has a positive coefficient of 6.519 with elderly people aged 60 years or older (p = 0.105), although this was relatively insignificant compared with the other groups. Next, we assessed how much financial inclusion differently affects financial inclusion depending on various personal characteristics through the IV model. Table 5 shows the results according to columns (7) to (12). Compulsory education in upper-secondary schools mostly affects having account ownership among women. For example, among women, owning an account rates would be around 30 percentage points higher if individuals completed their upper-secondary education. In contrast, among men, completion of upper-secondary education would raise the probability of having a financial account by 24 percentage points.
Table 5
OLS and IV estimations for account ownership and compulsory upper-secondary education.
| OLS regression | IV model |
| (1) Age 60+ | (2) Age 35–59 | (3) Female | (4) Male | (5) Poorest 40% | (6) Richest 60% | (7) Age 60+ | (8) Age 35–59 | (9) Female | (10) Male | (11) Poorest 40% | (12) Richest 60% | |
Constant | -76.918*** (20.090) | -68.444*** (17.496) | -84.374*** (18.302) | -52.234*** (17.858) | -92.018*** (19.808) | -54.468*** (16.907) | -71.033*** (22.170) | -63.113*** (19.442) | -78.505*** (20.537) | -47.478** (19.411) | -86.530*** (21.659) | -49.424*** (18.717) | |
ATM | 0.062 (0.043) | 0.064* (0.037) | 0.088** (0.039) | 0.062* (0.038) | 0.083* (0.042) | 0.068* (0.036) | 0.003 (0.051) | 0.010 (0.045) | 0.029 (0.047) | 0.015 (0.045) | 0.028 (0.050) | 0.017 (0.043) | |
Bank | 0.207* (0.119) | 0.208** (0.103) | 0.189* (0.108) | 0.116 (0.106) | 0.146 (0.117) | 0.145 (0.100) | 0.149 (0.132) | 0.155 (0.116) | 0.131 (0.122) | 0.068 (0.116) | 0.092 (0.129) | 0.096 (0.112) | |
GDP | 13.711*** (2.155) | 12.926*** (1.876) | 13.449*** (1.963) | 11.683*** (1.915) | 14.088*** (2.124) | 11.857*** (1.813) | 11.639*** (2.477) | 11.049*** (2.172) | 11.382*** (2.294) | 10.008*** (2.168) | 12.156*** (2.420) | 10.081*** (2.091) | |
Internet | 0.0001 (0.0001) | 0.0002 (0.0001) | 0.0003** (0.0001) | 0.0002 (0.0001) | 0.0003** (0.0001) | 0.0001 (0.0001) | 0.0001 (0.0002) | 0.0001 (0.0001) | 0.0002 (0.0002) | 0.0001 (0.0001) | 0.0003* (0.0002) | 0.0001 (0.0001) | |
Rural | 0.150 (0.104) | 0.203** (0.090) | 0.282*** (0.095) | 0.132 (0.092) | 0.257** (0.102) | 0.169* (0.087) | 0.262** (0.121) | 0.306*** (0.106) | 0.394*** (0.112) | 0.223** (0.106) | 0.362*** (0.118) | 0.266** (0.102) | |
Upper Secondary | 6.519 (3.992) | 7.801** (3.476) | 8.330** (3.636) | 6.380* (3.548) | 6.732* (3.936) | 7.570** (3.359) | 28.014*** (8.735) | 27.273*** (7.660) | 29.764*** (8.092) | 23.752*** (7.648) | 26.777*** (8.534) | 25.992*** (7.374) | |
Weak instruments | | | | | | | 23.425*** | 23.425*** | 23.43*** | 23.425*** | 23.425*** | 23.425*** | |
Wu-Hausman | | | | | | | 10.433*** | 11.357*** | 12.69*** | 8.513*** | 9.260*** | 10.850*** | |
Sargan test | | | | | | | 0.007 ,p = 0.932 | 1.282 ,p = 0.258 | 1.88 ,p = 0.170 | 1.482 ,p = 0.224 | 0.932 ,p = 0.334 | 2.086 ,p = 0.149 | |
Adj. R-squared | 0.696 | 0.736 | 0.739 | 0.689 | 0.715 | 0.719 | 0.633 | 0.677 | 0.674 | 0.636 | 0.662 | 0.659 | |
Observations | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | |
Notes
***, **, * denote significance at the 1%, 5%, and 10% level, respectively. Robust standard errors are in parentheses in all the models.
Table 6
OLS and IV estimations for account ownership and compulsory lower-secondary education.
| OLS regression | IV model |
| (1) Age 60+ | (2) Age 35–59 | (3) Female | (4) Male | (5) Poorest 40% | (6) Richest 60% | (7) Age 60+ | (8) Age 35–59 | (9) Female | (10) Male | (11) Poorest 40% | (12) Richest 60% |
Constant | -80.439*** (20.120) | -67.733*** (20.990) | -88.199*** (18.503) | -55.427*** (17.938) | -94.701*** (19.978) | -58.419*** (16.987) | -86.766*** (22.326) | -73.509*** (22.799) | -96.365*** (22.240) | -61.781*** (20.378) | -102.404*** (23.144) | -65.128*** (19.798) |
ATM | 0.079* (0.041) | 0.097** (0.043) | 0.110*** (0.038) | 0.079** (0.037) | 0.101** (0.041) | 0.088** (0.035) | 0.077* (0.046) | 0.095** (0.046) | 0.107** (0.045) | 0.077* (0.042) | 0.098** (0.047) | 0.086** (0.040) |
Bank | 0.205* (0.119) | -0.058 (0.124) | 0.194* (0.110) | 0.117 (0.106) | 0.155 (0.118) | 0.145 (0.101) | 0.135 (0.134) | -0.122 (0.137) | 0.103 (0.134) | 0.046 (0.122) | 0.069 (0.139) | 0.070 (0.119) |
GDP | 13.786*** (2.154) | 11.625*** (2.247) | 13.759*** (1.981) | 11.837*** (1.920) | 14.470*** (2.139) | 11.989*** (1.819) | 11.769*** (2.495) | 9.783*** (2.548) | 11.156*** (2.485) | 9.811*** (2.277) | 12.014*** (2.586) | 9.849*** (2.212) |
Internet | 0.0002 (0.0002) | 0.0005*** (0.0002) | 0.0003** (0.0001) | 0.0002 (0.0001) | 0.0003** (0.0001) | 0.0002 (0.0001) | 0.0002 (0.0002) | 0.001*** (0.0002) | 0.0004** (0.0002) | 0.0003* (0.0002) | 0.0005*** (0.0002) | 0.0003* (0.0002) |
Rural | 0.124 (0.102) | 0.190* (0.106) | 0.245*** (0.094) | 0.106 (0.091) | 0.226** (0.101) | 0.139 (0.086) | 0.155 (0.113) | 0.218* (0.116) | 0.286** (0.113) | 0.137 (0.103) | 0.263** (0.117) | 0.172* (0.100) |
Lower Secondary | 7.145 (4.721) | 3.850 (4.925) | 6.356 (4.341) | 5.951 (4.209) | 3.457 (4.687) | 7.732* (3.986) | 33.190*** (11.091) | 27.626** (11.326) | 39.973*** (11.048) | 32.107*** (10.123) | 35.165*** (11.497) | 35.348*** (9.835) |
Weak instruments | | | | | | | 19.665*** | 19.665*** | 19.665*** | 19.665*** | 19.665*** | 19.66*** |
Wu-Hausman | | | | | | | 9.111** | 6.871** | 19.167*** | 11.769*** | 14.165*** | 14.94*** |
Sargan test | | | | | | | 1.285 ,p = 0.257 | 0.075 ,p = 0.784 | 0.013 ,p = 0.908 | 0.002 ,p = 0.962 | 0.077 ,p = 0.781 | 0.00 ,p = 0.999 |
Adj. R-squared | 0.695 | 0.600 | 0.733 | 0.687 | 0.710 | 0.717 | 0.629 | 0.534 | 0.619 | 0.600 | 0.615 | 0.620 |
Observations | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 |
Notes
***, **, * denote significance at the 1%, 5%, and 10% level, respectively. Robust standard errors are in parentheses in all the models.
We also assumed that the required level of education differs for each sociodemographic characteristic and hence performed the same model with the dependent variable of the lower-secondary variable instead of upper-secondary education.
In Table 6, columns (1) to (6) of the OLS regression show that compulsory lower-secondary education had no significant correlation with financial inclusion except in the high-income class. In the high-income class, compulsory lower-secondary education and account ownership rates were significantly positive (b = 7.732, p = 0.054). Hence, in the high-income class, it was possible to analyze the effect of lower-secondary education based on financial inclusion through the IV model. The corresponding model can be checked through column (12). Compulsory lower-secondary education increased the rate of account ownership in the high-income class by approximately 35%. It showed the possibility of promoting financial inclusion with compulsory lower-secondary education in the high-income class.
Finally, we tested whether our results would still hold with excluding and including control variables. Columns (1) to (3) in Table 7 indicate the results of our models omitting ATM variable, columns (4) to (6) and (7) to (9) present the results without Bank and Internet variable, respectively. According to IV regression results, the coefficient of the influence of upper-secondary education on financial inclusion − 65.169 and − 73.529, respectively. We found that the results are quantitatively similar to those presented in Table 4, which showed the effect of compulsory upper-secondary education on financial account ownership.
The OLS and IV regression results with different sets of the control variables are in Table 8. Again, our results showed that the impact of compulsory upper-secondary education on financial inclusion is significantly positive. Using information about population growth rate and labor participation, respectively, for a robustness check through column (9), it was found that there is not much difference in the coefficient and significance level of upper-secondary education compared to the results without these variables. In all, these robustness checks further support our basic conclusion that compulsory upper-secondary education has a positive and significant effect on financial inclusion.
Table 7
OLS and IV estimations for account ownership omitting control variables.
| Excluded variable – ATM | Excluded variable – Bank | Excluded variable – Internet |
| (1) OLS | (2) First stage | (3) IV model | (4) OLS | (5) First stage | (6) IV model | (7) OLS | (8) First stage | (9) IV model |
Constant | -80.665*** (16.776) | -1.172** (0.371) | -67.224*** (18.908) | -71.037*** (17.574) | -0.872** (0.396) | -65.169*** (19.456) | -80.480*** (16.236) | -0.956** (0.382) | -73.529** (18.006) |
ATM | | | | 0.085** (0.037) | 0.002** (0.001) | 0.029 (0.045) | 0.071* (0.037) | 0.002** (0.001) | 0.019 (0.045) |
Bank | 0.191* (0.102) | 0.004* (0.002) | 0.106 (0.116) | | | | 0.157 (0.104) | 0.003 (0.002) | 0.104 (0.116) |
GDP | 14.173*** (1.754) | 0.032 (0.042) | 11.265*** (2.149) | 13.074*** (1.8873) | 0.004 (0.044) | 11.096*** (2.174) | 14.010*** (1.716) | 0.012 (0.042) | 11.978*** (2.017) |
Internet | 0.0002 (0.0001) | 0.0000 (0.0000) | 0.0002 (0.0001) | 0.0002* (0.0001) | 0.0000 (0.0000) | 0.0002 (0.0001) | | | |
Rural | 0.229** (0.091) | -0.003 (0.002) | 0.313*** (0.103) | 0.208** (0.091) | -0.003 (0.002) | 0.307*** (0.106) | 0.232** (0.090) | -0.003 (0.002) | 0.327*** (0.104) |
Upper Secondary | 9.066*** (3.406) | | 27.158*** (7.050) | 7.732** (3.485) | | 26.796*** (7.664) | 7.548** (3.504) | | 26.484*** (7.706) |
Expected Years | | 0.111*** (0.015) | | | 0.105*** (0.016) | | | 0.105*** (0.015) | |
Rights | | -0.056** (0.022) | | | -0.046** (0.022) | | | -0.045** (0.021) | |
Weak instruments | | | 27.232*** | | | 23.426*** | | | 23.282*** |
Wu-Hausman | | | 11.822*** | | | 10.701** | | | 10.359** |
Sargan test | | | 1.547 ,p = 0.2136 | | | 1.957 ,p = 0.1619 | | | 2.469 ,p = 0.116 |
Adj. R-squared | 0.723 | 0.650 | 0.667 | 0.727 | 0.657 | 0.669 | 0.726 | 0.659 | 0.669 |
Observations | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 |
Notes
***, **, * denote significance at the 1%, 5%, and 10% level, respectively. Robust standard errors are in parentheses in all the models.
Table 8
OLS and IV estimations for account ownership with further control variables.
| Additional variable – Lower Secondary | Additional variable – Population Growth | Additional variable – Labour Participation |
| (1) OLS | (2) First stage | (3) IV model | (1) OLS | (2) First stage | (3) IV model | (1) OLS | (2) First stage | (3) IV model |
Constant | -70.789*** (17.521) | -0.854** (0.384) | -66.320*** (18.860) | -72.572*** (18.437) | -0.694 (0.421) | -72.721*** (20.612) | -78.812*** (23.445) | -0.684 (0.493) | -67.334** (26.009) |
ATM | 0.074** (0.037) | 0.001* (0.001) | 0.030 (0.043) | 0.076** (0.038) | 0.001* (0.001) | 0.025 (0.046) | 0.067* (0.039) | 0.002** (0.001) | 0.021 (0.046) |
Bank | 0.131 (0.104) | 0.003 (0.002) | 0.089 (0.113) | 0.155 (0.105) | 0.002 (0.002) | 0.116 (0.118) | 0.160 (0.106) | 0.002 (0.002) | 0.101 (0.119) |
GDP | 12.286*** (1.901) | -0.006 (0.043) | 10.792*** (2.108) | 12.898*** (1.910) | -0.006 (0.044) | 11.246*** (2.208) | 12.940*** (1.919) | -0.001 (0.044) | 11.025*** (2.213) |
Internet | 0.0002* (0.0001) | 0.0000 (0.0000) | 0.0002 (0.0001) | 0.0002* (0.0001) | 0.0000 (0.0000) | 0.0002 (0.0001) | 0.0002 (0.0001) | 0.0000 (0.0000) | 0.0002 (0.0001) |
Rural | 0.211** (0.091) | -0.003* (0.002) | 0.294*** (0.102) | 0.215** (0.093) | -0.004* (0.002) | 0.341*** (0.112) | 0.207** (0.091) | -0.003 (0.002) | 0.302*** (0.105) |
Lower Secondary | 5.825 (4.106) | -0.292*** (0.095) | 5.338 (4.407) | | | | | | |
Upper Secondary | 7.156** (3.476) | | 23.046*** (6.814) | 7.651** (3.551) | | 28.584*** (8.171) | 7.115** (3.507) | | 25.580*** (7.689) |
Expected Years | | 0.129*** (0.017) | | | 0.102*** (0.016) | | | 0.107*** (0.016) | |
Rights | | -0.050** (0.022) | | | -0.047** (0.022) | | | -0.060** (0.026) | |
Population Growth | | | | 0.652 (1.110) | -0.028 (0.023) | 1.802 (1.302) | | | |
Labour Participation | | | | | | | 0.499 (0.816) | -0.013 (0.020) | 0.159 (0.902) |
Weak instruments | | | 29.488*** | | | 21.329*** | | | 22.935*** |
Wu-Hausman | | | 9.472*** | | | 11.556*** | | | 9.800*** |
Sargan test | | | 1.242 ,p = 0.2652 | | | 1.317 ,p = 0.2512 | | | 2.458 ,p = 0.117 |
Adj. R-squared | 0.731 | 0.678 | 0.748 | 0.727 | 0.659 | 0.659 | 0.727 | 0.657 | 0.673 |
Observations | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 | 147 |
Notes
***, **, * denote significance at the 1%, 5%, and 10% level, respectively. Robust standard errors are in parentheses in all the models.