Robustly Estimating Japan's Gini Coefficient for Individual Earned Income Using Household Survey and Tax Agency Data

This study investigates the benefits and caveats of using tax agency data through a descriptive analysis, and estimates the robust Gini coefficient for individuals' earned income by applying statistical tools combining household survey and tax agency data. We show that the advantage of the tax agency data is that it captures top incomes, while its weak coverage of female non-regular workers can be complemented by combining it with the household survey data. Further, the descriptive results show that the Gini coefficient computed using the household survey is larger than that from using tax agency data, despite not covering top incomes. This indicates that capturing the distribution of the middle and low incomes is more important to estimate the inequality level than capturing the top incomes in Japan. Moreover, the robust estimate of the Gini coefficient indicates that combining top incomes does not substantially affect the overall Gini index computed solely from the household survey data, which is distinct from the results for other countries in the literature. However, when we decompose the Gini coefficient into between- and within-group components of gender and employment status, combining the tax agency and household survey data is important. Although both data show an increase in the between-group component from 2014 to 2019, the integrated data indicate that the between-group contribution actually decreases from 2014 to 2019, reflecting the increases in the incomes of regular female workers.