Credit card interest rates don't generally relate to default probabilities in a sane way, as Mike Konczal explains here: http://rortybomb.wordpress.com/2009/05/

Credit card interest rates don't generally relate to default probabilities in a sane way, as Mike Konczal explains here: http://rortybomb.wordpress.com/2009/05/22/credit-card-reform-does-my-credit-cards-interest-rate-mean-anything/ http://rortybomb.wordpress.com/2009/05/22/credit-card-reform... There is no rational model that relates your credit card interest rate in a reasonable way to your own actions. If you send in a $5 payment two days late, that's often treated the same as sending in a $5000 payment one month late. But any sane bank officer will tell you that these two events are vastly different when it comes to default probabilities. Credit cards in the US are an insane debt instrument, and ideally, people would use them only for providing transactional credit and not for loans. Its funny, given the wealth of data available to credit card providers, they have the ability to carefully fine tune rates to a decent estimate of loss probabilities. If that happened, you'd see rates smoothly transition along a curve in response to your actions rather than see-sawing between two or three rates. But pricing in default risk isn't the goal since defaults are great for credit card companies. The more you default, the more fees and interest they can tack on, and many borrowers will end up paying most of that back. It is so much more profitable to earn $5000 on a debt of $1000 after all, even if you have to wait 2 or 3 years to actually get it.