The Korean Economic Forum
Problems and Corrections of the Household Income and Expenditure Survey
Nak Nyeon Kim (Dongguk University)Year 2022Vol. 15No. 1
Since the Household Income and Expenditure Survey is designed as rotation sampling in which a certain percentage of the sample is replaced every month, the number of months each sample is surveyed in a year spans 1-12 months (1-6 months after 2020). Statistics Korea calculates quarterly or annual statistics by simply averaging the survey results for the month from monthly data. This is fine for items with similar recurring monthly expenditures, but distortion occurs for items that do not. Depending on the number of months surveyed and the frequency of purchase of items, the annual statistics tend to be exaggerated by up to 12 times (three times in the case of quarterly statistics) than the actual ones, while items not purchased in the survey month are completely omitted from the survey. When calculating the overall average, it was found that such overestimation and underestimation due to omission offset each other. However, the average per household is not, so the consumption gap between households is wider than it really is, and the distortion has grown bigger in recent years. In this paper, the bias of items consumed by each household included in the annual data for 2015-16 and 2019-20 was corrected in a reasonable way, and the results were compared with the existing micro data.