가계조사의 행정자료에 의한 보정: 2016년 가계금융복지조사를 중심으로

Abstract

This paper compares survey results of the 2016 Survey of Household Finance and Living Conditions with the revised results using administrative data to reveal what biases respondents have in reporting their income and assets. First, in the case of earned income, the higher the income, the greater the tendency to underreport. In the case of financial income, many respondents did not report at all, and even if they responded, the income was greatly reduced. On the other hand, when income is low, the reported income tends to be larger than the income found in administrative data, and the gap between them is often very large. Factors such as misunderstanding or confusion among respondents regarding the concept or classification of income may have been at work. Second, non-sampling errors were mostly corrected by using administrative data. However, when compared with the tax data again, there were many omissions at the top, especially in the case of financial income. This is a sampling error where household surveys do not properly reflect the actual distribution. Third, in the case of assets, corrections were not made using administrative data, so the survey results were directly compared with the tax data. In the case of housing assets, the asset distributions of the two data were very close. In the case of land where the asset is more concentrated on the top floor, some of the top floors were missing, and in financial assets the problem was even worse.