I posted awhile ago on the mortgage crisis and the performativity of economic models and ratings agencies. In that post, i made the argument (via proxies like the NYT) that the models used at the major credit rating agencies (and, in particular, a faulty independence assumption) were at least partly to blame for creating the subprime mortgage mess. Another ratings agency-related explanation also makes sense, probably contributed at least somewhat to the crisis, and is a bit simpler:

Study Finds Flawed Practices at Ratings Firms – NYTimes.com:

The analyst at the credit ratings agency was blunt: “Let’s hope we are all wealthy and retired by the time this house of cards falters.”

That candid assessment, sent by e-mail to a colleague in December 2006, referred to the market for certain investments linked to subprime mortgages – investments that were assigned top AAA ratings from major agencies, only to later plummet in value.

That e-mail message and dozens like it were disclosed Tuesday in a blistering 37-page report issued by the Securities and Exchange Commission, which confirmed what many on Wall Street had long suspected: the major ratings firms, including Fitch, Moody’s and Standard & Poor’s, flouted conflict of interest guidelines and considered their own profits when rating securities, among other suspect practices.
…
The agencies continued to issue ratings despite frequent complaints from managers that they had neither the time nor manpower to measure the safety of investments sufficiently.

“We do not have the resources to support what we are doing now,” a managing analyst wrote in an e-mail message in February 2007.
…
The agencies also considered changing their ratings criteria to better compete with their rivals. “We are meeting with your group this week to discuss adjusting criteria for rating C.D.O.’s of real estate assets this week because of the ongoing threat of losing deals,” a business manager wrote in an August 2004 e-mail message, referring to collateralized debt obligations.

So, the combination of bad models and a bit of Williamsonian “self-interest seeking with guile” under bounded rationality?