We find that 75 percent of those balances had been paid off or charged off by February 2013. Charge-offs played a much smaller role in balance reduction than did paydown: 27.8 percent of balances were charged off, while 72.2 percent were paid down. Charge-offs accounted for a much larger share of balance reduction in the riskiest quintile and almost none of the reduction in the least risky quintile. After stratifying by risk score as of March 2009, balance, and utilization, we find that highly utilized accounts were no more likely to reduce their debt balances than low utilized accounts. We also find that low-utilization accounts were more likely to reduce their balances by paydown rather than default. By comparing accounts affected by unfavorable events, such as closures, freezes, and rate increases, with those accounts that did not experience such an event, we find that, while the aggregate results appear to be similar, there is a high degree of variation within risk quartiles, both with respect to debt reduction rates and to balance reduction shares attributed to paydown and to charge-off.