This study aims to construct household information in various ways by using individual-level data from the administrative records of the Health Insurance Service, and to analyze economic and social disparities between income groups.
The data used in this study is per the customized income-property DB built based on the National Health Information DB of the National Health Insurance Service. This data contains individuallevel information of the entire population (57.21 million people) for 13 years(2009~2021).
In this study, households are organized in five packages per the information on individual resident registration, health insurance, family relationship, and income activities of the customized income-property DB. In the five proposals, households are made up of families (1) who have same Resident Registration Household ID + NHIS dependents(H1), (2) who have same Resident Registration Household ID + non-cohabiting family members(H2), (3) who have same Resident Registration Household ID + disguised separated households(H3), (4) who have same Resident Registration Household ID(H4), and (5) who have same NHIS card ID(H5), respectively.
As a result of the construction of households, as of 2021, H1 had the smallest number of households with 21.21 million households, and H5 had the largest number of households with 27.62 million households. There was no significant difference between H2~H4 households with 23.45~23.95 million households. The proportion of elderly household owners over 65 years old was relatively low within H1 and H5.
Chapter 3 analyzes the income distribution of customized income-property databases and compares the result with the Household Financial Welfare Survey Data. First of all, comparing the distribution of personal income found that the income capture rate of the household financial welfare survey was generally higher than that of the customized income-property DB. The proportion of individuals with earned income was higher in the Household Financial Welfare Survey, and the average of individual's public pension was also higher in the Household Financial Welfare Survey. In terms of household income distribution, H2, H3 and H4 were quite similar. But significant differences were observed between the two groups, H2, H3, H4 and H1, H5. Compared with H2, H3, and H4, the household income level of the middle- and low-income groups was relatively high in H1, and the household income level of the middle-and high-income groups was lower in H5. Regarding the equivalized gross household income, the Gini coefficient and poverty rate of the Household Financial Welfare Survey were 0.396 and 21.1%, respectively. The Gini coefficient and poverty rate of the customized income-property DB were 0.486~0.537 and 28.1~32.3% depending on the household composition methods. The poverty rate of the elderly in H1 and H5 were 32.1% and 32.9% respectively, which were much lower than the 56.6% in the Household Financial Welfare Survey.
When comparing the average household property, H1 was the highest and H5 was the lowest. The disparity between classes of property was also lowest in H1 and highest in H5. The difference between H2 and H3 was relatively small. The property gap by income classper the quintile ratio was the lowest in H1 and the highest in H4. The income disparity among property classes measured by the Palma ratio was relatively high in H2 and H4, and lowest in H5.
In chapter 6, we derived two income-property combination indices reflecting the economic solvency of households, by applying two methods: property-annuity method and the tariff income method. It was found that the average characteristics of households do not vary significantly by the household formation methods. When comparing the case of considering income only and the case of considering both income and property, there were differences in the age of the household head and the presence of elderly household members. However, depending on how property was converted into and combined with income, the average characteristics did not change much.
On the other hand, the consistency between the indices was not high. We found that, depending on whether the property- annuity method or tariff income method is used, the identification of the lower or upper households is quite different. This means that it should be carefully considered in policy design how to define the solvency of households.
In Chapter 7, the explanatory power of the 100 percentiles of equilivalized income for the mortality rate was compared among the five household formation methods. And we also compared the five methods in terms of the relative mortality ratio of 10 income deciles. The explanatory power of the income percentile for mortality was the lowest in H5 and H1, while H2, H3 and H4 showed similar levels. Similarly, changes in the relative mortality rate ratio in the income decile showed the same pattern in H1 and H5, and the other three showed the same pattern. In H1 and H2, the mortality rate for both men and women decreased in line with the income increament. But in H3, H4, and H5, the mortality rate decreased sharply only in the top decile regardless of the gender.
In this study, we tried to propose a household composition methods by using the personal information in the customized income-property DB. If this DB could be linked with the Population Census Data of KOSTAT, it would save a lot of effort and time for household formations and accuracy of the data. It is necessary to improve the system so that these two data can be easily linked.
If more detailed address information is available in the customized income-property DB, and, if information on other items not included in the income and property can be added or linked, richer analysis will be obtainable. As the specificity of various factors of households has varied according to the changes in the economic and social environment, it is necessary to continue theoretical researches on the concept of household in addition to the supplementing data.
