Research

Linking 1.5 million delivery orders to households' daily credit position, we find that a household's cash use rises in a convex pattern with its credit utilization. This pattern reflects a motive to preserve available credit for future expenditure shocks that may go on credit. As utilization rises, the remaining credit is more likely to fall short of such a shock; therefore, each dollar of anticipated consumption paid on credit costs more, and households substitute cash for credit. Using a regression discontinuity design, we show that bank-initiated limit increases reduce cash use, most among households near their limit. We estimate that for a household that pays no interest, a dollar on credit carries a shadow cost of 5 cents at 20 percent utilization and 24 cents at 80 percent, close to the 25 cents of lifetime interest on a dollar of revolving debt.

We study the usage and welfare implications of earned wage access (EWA) using administrative and survey data from a Mexican FinTech provider. Adoption is significant, and usage concentrates at the end of the pay cycle. We develop a stationary buffer-stock framework, calibrated to the data, in which EWA provides liquidity insurance and consumption-timing alignment. Access is worth about 2.5% of a paycheck per year for the average user, and an order of magnitude more for impatient, low-liquidity workers. Automatic repayment makes these gains more robust to present bias than with revolving credit. EWA usage is associated with higher employee retention, as the framework implies.

How resilient are banks to local funding shocks? We exploit a segment of Chinese wealth management products (WMPs) funded locally but backed by national assets, allowing us to isolate banks' strategic responses to local funding pressures. Using city-level inflation to characterize investor demand shifts, we find that banks pass through approximately 60% of these shocks to WMP yields. Large, geographically diversified banks respond more actively, by raising yields and contracting issuance in shock-hit markets more sharply than smaller banks. They then hedge by expanding the issuance of lower-yield WMPs in unaffected regions. Our findings reveal that for large banks, geographic diversification is not a tool for passive insulation but a mechanism enabling active, high-frequency liability management through their internal capital markets.