Credit ratings of Chinese households using factor scores and K-means clustering method

Abstract

There is a long line of literature on credit ratings of financial institutions, enterprises and individuals, while less attention has been paid to rating credits of the family unit. Considering that Chinese people have a special sense of family, it is more important to evaluate credit ratings of Chinese households. Using a data sample of 38,789 households from the China Household Finance Survey (CHFS), this paper collects 15 indicators of credit rating according to Sesame Credit and the US FICO credit system, including seven qualitative indicators (education, marriage, health, work unit, insurance and provident fund, agriculture, industrial and commercial operations and real estate status), seven quantitative indicators (relatively stable income, investment income, survival debts, proportion of real estate, proportion of household M2, survival expenditure and consumption structure upgrade expenditure) and one interval index (age). Initially, a K-Means clustering method using a revised measure of household financial fragility is used to rate the credits of Chinese households. This method can well distinguish households of different financial fragilities. However, this simple reclassification leads to an “inverted pyramid” problem (households with high credit rating are more than those with low credit rating). Therefore, an augmented credit rating method is developed to reclassify Chinese households, which integrates factor analysis and K-Means clustering using financial fragility, and families with differing household financial fragility can be clearly divided into different credit levels.