Factors and Risks of Household Over-indebtedness Using a New Measure based on Conditional Quantiles (in Korean)

This paper suggests a new measure of household over-indebtedness based on the concept of conditional quantiles, and use the measure to examine the factors and the risks of the over-indebtedness. We first consider a distribution of household debt conditional on the overall payment ability, and specify the over-indebted household as the one whose debt exceeds θ-quantile of the conditional distribution. The overall payment ability is set to cover a variety of factors such as disposable income, assets, expected income, and consumption backgrounds. θ-quantile is estimated using Chernozhukov and Hong (2002)'s three step estimator for censored quantile regression. The empirical analysis shows a clear evidence of heterogeneous behavior of the over-indebted households compared to the others. We find that over-indebted households' income and real estate elasticities of debt are significantly higher and their debt/income ratio has increased faster in recent periods of low interest rates. Simulation study shows that these distinctive properties cause over-indebted households facing far greater default risks under macroeconomic shocks such as interest rate hikes and house price drops. We also find that lower-income and elderly-headed households' over-indebtedness problem is more serious in that they have significantly higher debt/income ratio compared to the other age and income groups.