Abstract
This paper studies the role of time-varying credit limits through the lens of a life cycle incomplete markets model calibrated for the U.S. Changes in credit card limits are explained by observable household characteristics, and the estimated unobservable variation is quite large. The quantitative exercise shows that even though young households are more indebted in an economy with stochastic borrowing limits, aggregate consumption is not greatly affected by transitory or persistent shocks of this type. However, in the presence of these shocks households lose part of their ability to self-insure against other uninsurable idiosyncratic shocks, such as labor income shocks. Decomposing this loss, I find that it operates mainly through the asset-distribution channel — the effect of stochastic limits on precautionary wealth — rather than through the direct response of the consumption policy function. These conclusions are robust to a reduced-form treatment of default, which if anything attenuates the estimated effects.