Digital Financial Capability and Household Wealth Portfolio Diversification in China: Evidence from a CHFS-Derived Balanced Panel

Abstract

Objectives. This study examines whether within-household changes in digital financial capability are associated with changes in household wealth portfolio diversification in China. Data and Methods. We analyse a publicly released, unweighted balanced panel derived from the 2017 and 2019 China Household Finance Survey, comprising 18,396 household-year observations for 9,198 households. Diversification is measured by one minus the Herfindahl–Hirschman index across five asset categories, asset count, and normalised entropy. Household fixed-effects models include observed time-varying controls and a survey-year effect, with standard errors clustered by household. Results. The fixed-effects coefficient for digital financial capability on one-minus-HHI is 0.3476 (95% confidence interval from 0.2617 to 0.4335). A one-standard-deviation increase in capability corresponds to an approximately 0.030 increase, or 8.6% of the outcome mean. A household bootstrap with 500 replicates gives a percentile interval from 0.2478 to 0.4328. Positive associations also appear for asset count and normalised entropy, and remain under strict original-ratio and asset-count threshold checks. Component regressions associate higher capability with a lower housing share and higher productive, non-risky financial, and risky financial shares. Rural and income interactions are not statistically distinguishable from zero. Conclusions. Digital financial capability is positively associated with portfolio diversification within this released two-wave sample. The design does not identify a causal effect and, because survey weights and city identifiers are unavailable, does not support population-representative or subnational claims.