Methods
Our study uses a sample from The Financial Literacy Survey 2016 conducted by the Central Council for Financial Services Information (Bank of Japan). The Financial Literacy Survey 2016 was an internet survey conducted in February and March 2016 to investigate the level of financial literacy among Japanese people. The sample comprises 25,000 individuals between 18 and 79 years old from all prefectures in proportion to the Japanese demographic structure. Ethical approval was not required since this study was based on secondary analysis of publicly available data.
Definitions of variables
Dependent variables
Financial literacy denotes not merely financial understanding (i.e. financial knowledge and education), but also practical management of financial assets, which is considered helpful for individuals in accumulating assets [19,20]. A previous study suggests that there are three important components of financial literacy that focus on understanding [3]: (1) numeracy, or the capacity to calculate a simple compound interest rate; (2) understanding inflation; and (3) knowledge about stocks, stock mutual funds, and risk diversification.
For our analysis, we measured knowledge-based financial literacy by the number of correct answers out of 25 questions on the survey (Table 1). The items measure multiple dimensions of financial literacy as they contain a wide range of topics (e.g. financial knowledge and norms), including items used in previous studies [3,21,22].
Table. A list of questions to measure knowledge-based financial literacy. (See the original article for the full table.)
In addition to measuring financial literacy, we assess the association between age and self-rated financial literacy to evaluate how our definition of financial literacy and self-rated financial literacy diverge with age, considering that previous literature indicates that people tend to overestimate their level of financial literacy [3]. Self-rated financial literacy is determined by a five-point Likert scale from ‘very high’ to ‘very low’ for the question, ‘What level is your general knowledge about finance compared with others?’
To complement the understanding of financial literacy measured mainly by financial knowledge, we focus additionally on financial behaviours and attitudes, which can contribute to positive outcomes of financial literacy [15,23]. Financial behaviour measures people’s action in a financial transaction, which can be derived from financial knowledge. In contrast, financial attitude, as asked in the survey, measures the individual’s time preference for the transaction. Those who put too much value on the present and less on the future may suffer from a shortage of financial assets in the long run due to a lack of appropriate financial planning and management. Thus, an excessive discount of future benefits can be regarded as a negative in financial decision making, at least for younger people.
Following the definitions in a previous study [15], we use the following qualitative questions to measure financial behaviours and attitudes.
Financial behaviour:
- Before I buy something I carefully consider whether I can afford it.
- I pay my bills on time.
- I set long term financial goals and strive to achieve them.
- I keep a close personal watch on my financial affairs.
Financial attitude:
- I find it more satisfying to spend money than to save it for the long term.
- I tend to live for today and let tomorrow take care of itself.
- If I had the choice of (1) receiving 100,000 yen now or (2) receiving 110,000 yen in 1 year, I would choose (1), provided that I can definitely receive the money.
Respondents are asked to rate each item on a five-point Likert scale, where 1 = agree and 5 = disagree. Aggregating the responses, we create scores to measure financial behaviours and attitudes; these are transformed so that higher values represent more preferable financial behaviours and attitudes, namely, premeditated financial considerations.
Independent variables
The explanatory variables comprise individual demographic and socioeconomic factors that affect financial literacy: the age of the respondent, the square of the age, educational attainment, occupation, financial education, household income, and financial assets: Education is categorised into three groups: high-school graduate or lower; vocational college, junior college or equivalent; and university graduate or higher. Occupation is divided into five categories: employee or public official, self-employed, part-time, not employed (including domestic worker), and others. Financial education is a binary dummy variable, which takes a value of one if a respondent has received any financial education (e.g. about life planning and management of a family budget) and zero otherwise. Household income denotes the gross annual household income from the previous year, including revenues from all assets. Household income was not equalised as the information on the number of household members was not available. The financial assets refer to the amount of financial assets (e.g. savings and stock) that a household has at the time of the survey.
Empirical strategy
Financial literacy and individual characteristics
We begin by investigating financial literacy, self-rated financial literacy, and age to test hypothesis 1. We use the ordinary least squares method for the analysis of financial literacy and the ordered logistic regression for self-rated financial literacy.
Gender differences in financial literacy
To test hypotheses 2 and 3 (gender gaps in financial literacy), we use the Blinder-Oaxaca decomposition [24,25]. The decomposition model is formulated as follows: where Yi denotes the financial literacy of individual i, Xji represents independent variables (e.g. age and education) for individual i, and εi is a stochastic disturbance. βj is the parameter to be estimated.
Assuming E (βj) = βj and E (εj) = 0, the estimations of the subsample means for women and men, respectively, are: F and M indicate women and men, respectively, and Y- and X- denote the sample means.
Thus, the difference in the sample means of men and women is given by:
Here, differences in financial literacy between men and women can be decomposed into three components: the endowment, coefficient, and interaction effects. The endowment effect denotes the difference in the distribution between men and women of the explanatory variables that affect financial literacy, while the coefficient effect refers to the gender difference in the responses of the explanatory variables to the outcome. The interaction effect captures the existence of both the endowment and coefficient effects, including differences in the intercept.
To test hypothesis 2, we use the significance of the coefficient of the female dummy variable in the model testing hypothesis 1. Next, we decompose the gender differences in financial literacy to test hypothesis 3 using the Blinder-Oaxaca decomposition model as shown in Eq (1).
For financial behaviours and attitudes, we adopt the same procedures as the estimates for financial literacy. To account for the fact that financial knowledge can affect financial behaviour and attitude, we include financial literacy as a covariate in the models as well.
We exclude students (n = 1,212) from our analyses, since they may not be financially independent. Among the remaining 23,788 respondents, there are some whose educational attainment is unknown (n = 37), some who did not score their own financial literacy (n = 614), and some who lacked income information (n = 4,204) or financial asset information (n = 3,705). Thus, the final usable number of complete cases equalled 15,228 individuals. Table 2 shows the descriptive statistics by gender. The proportions of employees/public officials and self-employed individuals are higher among men, while the proportions of part-time workers and those not working are higher among women. Furthermore, more men than women graduated from university or higher and received financial education.
Table. Descriptive statistics for complete cases. (See the original article for the full table.)
To address the issue of potential non-response bias, we adopt multiple imputation for missing variables by the missing at random assumption and predictive mean matching using variables of financial attitudes, behaviours, and knowledge, adding 20 imputations. All analyses are performed using Stata version 17.0.
CC BY 4.0. "Age, gender, and financial literacy in Japan" by Okamoto S, Komamura K, PloS one (2021). https://doi.org/10.1371/journal.pone.0259393. Licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Peer-reviewed and open access; full text from PubMed Central (PMC8598074). Headings and layout adapted for reading; figures and tables are in the original.