Compulsory education enhances financial inclusion across socioeconomic groups: a global analysis
Data
We used cross-country level data from various sources. First, data on national account ownership status, population information by country, GDP, and internet server possession information were collected from the WDI and Findex data of the World Bank. Data related to the physical accessibility of financial institutions were secured from the financial access survey of the IMF. If the information available was not from 2017, it was replaced by data from 2011–2016. To generate the instrumental variable for education, we collected education-related information by country from the UNDP in conjunction with the Human Rights Score from Our World in Data (Fariss, 2019; Schnakenberg and Fariss, 2014). Table 1 presents the data descriptions and sources in detail.
Our analysis targets were the 160 countries for which information was available from various sources. We reduced all observations higher than the 99th percentile of each variable to the value in account ownership rate of the population aged 15 or older. All values lower than the first percentile of each variable were modified similarly to ensure that the statistical results are not influenced by outliers. We deleted eight countries for which it was difficult to perform data analysis because of missing values and analyzed the final 147 countries. Supplementary Table S1 lists all the countries included in the study.
Dependent variable
To identify the level of financial inclusion, we chose account ownership rate as the dependent variable, based on socio-demographic characteristics such as age, gender, and income level in each country. As financial inclusion is an abstract concept, it is not easy to measure it quantitatively. According to Demirgüç-Kunt et al. (2018), owning an account is the most fundamental index and the most necessary factor to measure financial inclusion by country. Sahay et al. (2015) mentioned that financial account ownership is the first stage of financial inclusion. Demirgüç-Kunt et al. (2022) determined the level of financial inclusion using three dimensions: access to financial services, usage of financial service, and the quality of products and service delivery. Cámara and Tuesta (2014) developed a financial inclusion index with a three-dimensional approach integrating usage, barriers, and access to formal financial services. These three factors were evaluated through the following measures: (i) whether an individual possessed at least one financial product and had savings or loan experience; (ii) why an individual could not use a financial service; (iii) how many bank branches and automated teller machines (ATMs) were built per country. However, the index used in previous studies to measure accessibility is insufficient to reflect the recent financial market trends as mobile and internet banking services have emerged as new channels for accessing official financial services. In fact, financial technologies, including those that are intangible-driven and combine e-finance and internet technologies, promote sustainable development and reshape the banking industry (Fabregas and Yokossi, 2022). Thus, this study measured the financial inclusion level using individual account ownership rates at multiple institutions (e.g., banks, micro-finance institutions, regulated financial institutions, and mobile money providers).
Data were collected through a questionnaire survey conducted by the World Bank of people aged 15 or older. Account ownership rates were calculated from the proportion of the population holding accounts with banks or other financial institutions or personally using a mobile money service. The sample size was 147, the range of values was 0–100, and the sample mean was 59.69. We considered the account ownership rates of middle-aged (aged 35–59) and older adults (aged 60 or older) people to analyze the level of financial inclusion based on age. Moreover, we categorized the income level standard into the following: low-income, corresponding to the lower 40% of an economy’s households by income, and high-income, corresponding to the upper 60% of households by income within the population aged 15 or older. We have provided detailed descriptive statistics for all the dependent variables in section “Results”.
Explanatory variables
The major independent variables of this study were compulsory upper-secondary and lower-secondary education by country. We calculated dummy variable Upper Secondary as follows:
$${Upper}{{Secondary}}_{i}=\left\{\begin{array}{c}1{if\; given\; country}i{reported\; upper\; secondary}\,\\ \,\\ \,{education\; is\; legally\; required\; for\; all\; students}\\ \,\\ 0{otherwise}\,\end{array}\right.$$
(1)
The upper-secondary variable yields 1 or 0 depending on whether upper-secondary education is compulsory in a country. The lower-secondary variable is also binary, which is 1 if lower-secondary education is compulsory in the corresponding country and 0 otherwise.
Control variables
Several determinants might arise as significant factors to financial inclusion. These include economic development, population and its density, demographic factors, and physical infrastructure (e.g., electricity and internet networks). Importantly, financial inclusion is affected by both demand- and supply-side economics, however, demand-side factors have the greater influence (Ramakrishna and Trivedi, 2018). In this study, we controlled for economic growth in each country and supply-side factors, including banking penetration and accessibility, as they might influence the decision to use financial services. Hence, attention can be focused on the influence of sociodemographic characteristics on the demand-side. Indeed, accessibility to financial institutions is major factor of financial inclusion in many existing studies on drivers of financial inclusion (Cámara and Tuesta, 2014; Vo, 2024). According to Czech et al. (2024), conventional financial institutions have not provided or have failed in providing steady financial service for rural areas. Exclusion based on geographical factors makes financial inclusion in rural areas difficult (Wen et al. 2024). Hence, we assumed that geographical factors cause financial exclusion and used the rural population rate by country as another control variable. Moreover, as already verified in many research papers, a country’s growth index (e.g., GDP) has a positive effect on financial inclusion. GDP per capita was used as a proxy variable for economic growth to control this effect.
Method
We calculated descriptive statistics to check the status of financial account ownership depending on age, gender, education level, labor force participation, and income inequality. As Fig. 1 illustrates, education has the greatest divide in financial inclusion in comparison to all the other demographic characteristics. There is a great inequality of financial inclusion between those with secondary or higher education and those without. For example, as of 2017, inequality in financial exclusion based on age was ~5.1% with inequality of ~7.1% for gender, 15.8% for labor force participation, and 13.5% for income inequality. Subsequently, inequality based on education level showed the highest level of ~19.4%. According to the analysis, inequality in financial inclusion based on education level reached ~18.3%, 18.5%, and 19.4% in 2011, 2014, and 2017, respectively; therefore, it gradually increased.
Age shows the gap between individuals who report having a bank account and those who report not owning one for the age range of 35–59 and 60 and over; Gender indicates the extent of the bank account gap between males and females; Education Level measures this for individuals who have completed secondary education or more and primary education or less; Labor Force Participation measures this for individuals in and out of labor force; Income Inequality measures this for individuals with income falling under the richest 60% and the poorest 40%. Source: The authors’ analysis.
After analyzing account ownership rates by country, a classical ordinary least-squares (OLS) regression was performed to analyze the relationship between compulsory upper-secondary education and financial inclusion:
$$\begin{array}{c}{{Account}}_{i}=\alpha +{\beta }_{1}\times {AT}{M}_{i}+{\beta }_{2}\times {Ban}{k}_{i}+{\beta }_{3}\times {GD}{P}_{i}\\\qquad\qquad\qquad\qquad\qquad\qquad\quad +{\beta }_{4}\times {Interne}{t}_{i}+{\beta }_{5}\times {Rura}{l}_{i}+{\beta }_{6}\times {Upper\; Secondar}{y}_{i}+{\epsilon }_{i},\end{array}$$
(2)
where i and Account represent a given country and account ownership in each country, respectively. Proxies for accessibility to financial institutions and services are the number of commercial bank branches and ATMs. ATM and Bank denote the number for both ATMs and commercial bank branches per 100,000 persons. Internet indicates the number of secure internet servers per 1 million people. We also included GDP per capita in current US dollars as a proxy variable for economic growth. GDP, Rural, and Upper Secondary represent the logarithm of GDP per capita in current US dollars, the proportion of people living in rural areas of the total population, and a dummy variable equal to 1 if the upper-secondary level of education is compulsory for each country and 0 otherwise, respectively.
However, the regression results were insufficient for analyzing the causal relationship on financial inclusion. Specifically, compulsory upper-secondary education may be endogenous to the outcomes being studied; hence, any such variable bias should be eliminated. Upper-secondary education has the goal of cultivating students’ capability of receiving higher education or managing economic life in society. Moreover, students receiving upper-secondary education often attain the legal working age; consequently, they are likely to have both the right to education and to work (United Nations Educational, Scientific and Cultural Organization Institute for Statistics, 2016). Hence, students completing compulsory upper-secondary education are most likely to participate in economic life after graduation; thus, their probability of having accounts for wage receipt is naturally increased.
To minimize endogeneity bias, we applied the IV model widely used for removing variable bias (Wooldridge, 2002). We used the expected years of schooling by country and the human rights index as instruments. To ensure the effectiveness of the instruments that we defined in the IV model, the following two necessary conditions were applied: (i) the instrumental variables should be partially and sufficiently strongly correlated with compulsory secondary education as the regressor after controlling for the other independent variables; (ii) The corresponding instrumental variables and error terms in the linear equation should be uncorrelated.
Our instruments met these two conditions based on several criteria. First, although secondary education might have been included in the period needed for students to complete the given education, it may not have been included because of internal or external factors in the country. According to Indicators OECD (2012), many countries require people to complete their primary and lower-secondary education; however, upper-secondary education is optional in many countries in Africa and Asia. Thus, the expected years of schooling by country are partially related to the existence of compulsory upper-secondary education.
Second, although each country has a certain expected period of schooling, not everyone in the nation is required to attend school during those years of education. Students can optionally extend or suspend the education period separately from the schooling period (Azevedo et al., 2021). The expected period of schooling is simply the predictive index showing the period when a student is considered to belong to the education system in their childhood, and this period does not necessarily increase productivity or individual account ownership.
According to Lusardi and Mitchell (2011), financial literacy is not simply acquired through formal education but is significantly influenced by experiences with financial products, financial management, and knowledge gained through work. Additionally, studies related to financial literacy suggest that actually using financial products and managing money improves financial understanding. Particularly, experiences gained through employment or managing family finances are deemed crucial (Brown and Taylor, 2014; OECD, 2023). Hanushek and Wößmann (2007) established that quality of education is substantially more crucial for economic outcomes than the mere quantity of education (i.e., expected years of schooling). They found that spending more years in schooling (i.e., completing five or even 9 years of schooling) does not mean that students in typical developing countries have become functionally literate in basic cognitive skills. In other words, financial knowledge is more influenced by life experiences after completing at least compulsory education, rather than by the number of years spent in school. Expected Years cannot directly determine financial inclusion.
The human rights index affects compulsory education, and many recent studies depend on compulsory education laws as an exogenous proxy for education (Black et al., 2008; DeCicca and Krashinsky, 2020). Since the Second World War in the 20th century, it has been widely recognized that all children have the right to be educated through compulsory schooling, and many countries have established legal provisions to support the right to education (Rothbard, 1979). Meanwhile, there is no observed correlation between the human rights index and our dependent variable. Many studies indicate that social and economic factors affect financial inclusion (Bhatia and Dawar, 2023; Mishra et al., 2024), but the human rights index is not considered as a significant variable. Bain (2023) noted that in highly financialized states, financial system priorities often overshadow basic obligations such as public service provision and human rights protection. While some studies suggest that the political environment may influence financial knowledge, it is difficult to attribute this to the human rights index. Prete (2024) found that the political environment and government policies impact financial literacy, particularly in marginalized regions. Similarly, Pinto (2016) showed that countries with more stable political environments and higher governance quality tend to have higher levels of financial literacy, as trust in financial institutions is higher and broad educational initiatives are supported. These results stem from political environment and policy rather than the impact of the human rights index.
Lastly, we also verified the technical validity of the instrumental variables using the Wu-Hausman and Sargan tests.
We assumed that compulsory education positively affects the possibility of owning a financial account. An explicit two-stage model was developed, and the first stage involved a logistic regression with Upper Secondary on the other regressors and two instruments as follows:
$$\begin{array}{c}{UpperSecondar}{y}_{i}={\gamma }_{1}+{\gamma }_{2}\times {AT}{M}_{i}+{\gamma }_{3}\times {Ban}{k}_{i}+{\gamma }_{4}\times {GD}{P}_{i}+{\gamma }_{5}\times {Interne}{t}_{i}\\\qquad\qquad\qquad\qquad +{\gamma }_{6}\times {Rura}{l}_{i}+{\theta }_{1}\times {Expected\; Year}{s}_{i}+{\theta }_{2}\times {Right}{s}_{i}+{\epsilon }_{i}.\end{array}$$
(3)
To support the hypothesis that the minimum education level needed for financial inclusion will differ depending on demographic characteristics, the Lower Secondary variable was applied, instead of the Upper Secondary variable, with the other conditions set to be the same to construct the OLS regression and the IV model.
Finally, we conducted a robustness test by omitting control variables and applied combinations of expanded exploratory variables to verify the robustness of the relationship between compulsory education and financial inclusion. It is because an effective way of conducting a robustness test is to apply different measurement methods for various explanatory variables (Lyu et al., 2023).
link
