Good post today on Credit Slips, in which Adam Levitin discusses the problems with consumer credit, particularly the terms of credit cards that blur customers' real expectations. Of course, that is exactly the point, where "consumer cognitive issues" lead to huge profits for the credit card issuers.
A more general problem is the extent to which statistical models dominate certain industries and create asymmetric risk. A lot of people wonder how credit card companies can be so lax as to send offers to children, dead people, even dogs. The answer is that they don't care. There is a model somewhere that could be used to estimate, if you signed every person in America up for a credit card, how much they'd use the card, how likely they'd be to default, how big a balance they'd keep, and so forth. That model can be used to determine the terms of the card, the interest rate, and you can bet that there is a substantial premium built in.
What some people don't understand is the statistical nature of that; it is not that they can tell whether or not you will default. It is that, given a thousand you's, they know pretty well how many of you will default. The same is true of the well-known FICO score (also known as your credit score); no single score has to be accurate as long as the distribution of scores leads to big profits.
These models work only in the aggregate - for the individual, the decision is binary. You're denied the loan, maybe even the job (more companies are running a credit check on job applicants). The credit companies will do fine; even if they miscall a few, they'll just change the terms anyway, it's not as if you can really go anywhere (collusion vs. competitive pressure, please discuss). They've cut their risk to 0.
For you, of course, a billing dispute or lax delivery by the Post Office can cause you some serious problems - massive late fees or a permanent interest rate bump. Your risk is huge. Just another example of how the big people can shift the risks of business (and society) to the little people.
A more general problem is the extent to which statistical models dominate certain industries and create asymmetric risk. A lot of people wonder how credit card companies can be so lax as to send offers to children, dead people, even dogs. The answer is that they don't care. There is a model somewhere that could be used to estimate, if you signed every person in America up for a credit card, how much they'd use the card, how likely they'd be to default, how big a balance they'd keep, and so forth. That model can be used to determine the terms of the card, the interest rate, and you can bet that there is a substantial premium built in.
What some people don't understand is the statistical nature of that; it is not that they can tell whether or not you will default. It is that, given a thousand you's, they know pretty well how many of you will default. The same is true of the well-known FICO score (also known as your credit score); no single score has to be accurate as long as the distribution of scores leads to big profits.
These models work only in the aggregate - for the individual, the decision is binary. You're denied the loan, maybe even the job (more companies are running a credit check on job applicants). The credit companies will do fine; even if they miscall a few, they'll just change the terms anyway, it's not as if you can really go anywhere (collusion vs. competitive pressure, please discuss). They've cut their risk to 0.
For you, of course, a billing dispute or lax delivery by the Post Office can cause you some serious problems - massive late fees or a permanent interest rate bump. Your risk is huge. Just another example of how the big people can shift the risks of business (and society) to the little people.
No comments:
Post a Comment