Liquidity Constraint Threat and Household Indebtedness

Using proprietary high-frequency credit card data from Singapore, we study the causal effect of a public policy introducing strict credit constraint, on household consumption and debt repayment behavior. The new policy requires the card issuers to completely suspend any credit provided to heavily indebted individuals after two years of the law announcement. While the punishment for heavy indebtedness is very severe, the law is not binding for most of the population, given that it is hard to trigger the law before personal bankruptcy. We investigate the response of household debt, payment, and spending behavior to the law change during the announcement period and implementation period under a difference-in-difference setting. The revolvers, who tend to accumulate credit card debt without paying it off in time, are expected to be more affected by the credit suspension threat policy. The significant response of households even far away from triggering the law could arise from two mechanisms: consumers’ precautionary motive to avoid future troubles, or the spillover effect from those consumers that have been punished. We disentangle the channels from two perspectives: the effective time, and the affected consumer group. Specifically, the spillover effect channel shall only work after the implementation, and driven by peer consumers of heavily indebted individuals; while the precautionary motive channel can take effect immediately after the announcement of law, and does not suggest concentration effect in the peer consumers only. Then we extend to investigate whether the individuals are truly reducing their debt or simply shifting the debt to other accounts. Similar responses in both revolving and non-revolving account of revolvers suggest change in individual’s attitude towards revolving debt accumulation, otherwise it is more likely to be debt shifting. Given the rich demographic characteristics, we are able to further look into the heterogeneity effects. Specifically, we can divide consumers into subsamples by gender, ethnicity, marital status, age, home ownership, income, credit utilization, etc. With the transaction level credit card spending data, we can also look at the heterogeneity in spending types. In the last part, we shall focus on the falsification and robustness checks, such as PS matching, constructing pseudo events, etc. This study bears both academic and practical contributions. Academically, we contribute to literatures on consumer financial regulations, credit constraints, and precautionary motives. Practically, our study estimate the effectiveness of a financial policy targeting reducing household revolving debt, which can serve as good reference for policy makers.