> I have yet to hear any of them put actual numbers to their assertions. If you have any, I would honestly love to hear them.
As would I. Unfortunately, actual numbers seem hard to come by. Performance of CRA loans simply wasn't tracked. From a Fed study some years prior to the crisis:
We received responses from 143 of the 500 institutions to which we sent the survey (a 28.6 percent response rate). These responses and our follow-up telephone contacts revealed that banking institutions generally do not track profitability and performance separately for CRA-related lending... [1] [2]
For this reason, arguments absolving the CRA [3] instead look at high interest rate loans to low income households. If this is a good proxy for CRA lending, then CRA lending was a drop in the bucket.
However, there is evidence to the contrary. Edward Pinto, chief credit officer at Fannie Mae during the 1980s, published a report in 2008 that has some numbers you might be interested in. [4] From another article summarizing them [5]:
While it’s difficult to find data on how these CRA loans have performed, the few bank reports that break out these loans show much higher default rates than prime loans. Pinto’s research indicates that Fannie and Freddie’s affordable housing loans are a good proxy for CRA loans performance. For example, Fannie’s delinquency rate on its $900 billion in high-risk loans, 85 percent of which are affordable housing loans, was 11.36 percent at September 30, 2009—about 6.5 times the 1.8 percent delinquency rate on the GSEs’ traditionally underwritten loans. While the vast majority of CRA loans were fixed rate and did not have higher interest rates like other subprime loans, Pinto’s research has found that most had high risk characteristics such as small downpayments and impaired credit, signaled by a low borrower FICO score. As a result, loans with these high-risk characteristics have performed similarly to fixed rate subprime loans.
Perhaps more important in its implications for the financial crisis, Fannie and Freddie did not report that their mortgages were subprime and Alt-A. So instead of the 12 million prime loans market participants expected, Fannie and Freddie had added 12 million subprime and Alt-A loans to the market...
So, by the middle of 2008, there were almost 27 million subprime and Alt-A loans in the U.S. financial system. This amounted to almost 50 percent of all mortgages. More than two-thirds of these mortgages were held or guaranteed by government agencies like FHA (about 4.8 million), and the GSEs Fannie and Freddie (12 million loans), and by U.S. banks (a residual of about 2.2 million).
But I personally think the real answer is, we just don't know, CRA lending wasn't tracked. Hopefully next time we decide to experiment on this scale, we collect some data. ;)
The whole theory is bizzare. Banks didn't have to be forced to write stupid loans at the lead-in to the crisis. Once securitization made it easy to take a cut up front but sell them off onto someone else's books before they blew up, brokers and banks (including mortgage banks unaffected by the CRA) were eager to push through as many bad deals as they possibly could, with basically any buyer who could fog a mirror. Wall St. was so desperate for stuff to bet on that when the supply of borrowers ran low, they resorted to making new side bets on the outcomes of old loans (through synthetic CDOs made up of credit default swaps).
Blame the credit rating cartel for declaring large tranches of the whole mess to be "AAA investment grade" and thus marketable to huge institutions that had no business being involved, because they didn't think homebuyer defaults would be strongly correlated due to underlying forces.
As would I. Unfortunately, actual numbers seem hard to come by. Performance of CRA loans simply wasn't tracked. From a Fed study some years prior to the crisis:
We received responses from 143 of the 500 institutions to which we sent the survey (a 28.6 percent response rate). These responses and our follow-up telephone contacts revealed that banking institutions generally do not track profitability and performance separately for CRA-related lending... [1] [2]
For this reason, arguments absolving the CRA [3] instead look at high interest rate loans to low income households. If this is a good proxy for CRA lending, then CRA lending was a drop in the bucket.
However, there is evidence to the contrary. Edward Pinto, chief credit officer at Fannie Mae during the 1980s, published a report in 2008 that has some numbers you might be interested in. [4] From another article summarizing them [5]:
While it’s difficult to find data on how these CRA loans have performed, the few bank reports that break out these loans show much higher default rates than prime loans. Pinto’s research indicates that Fannie and Freddie’s affordable housing loans are a good proxy for CRA loans performance. For example, Fannie’s delinquency rate on its $900 billion in high-risk loans, 85 percent of which are affordable housing loans, was 11.36 percent at September 30, 2009—about 6.5 times the 1.8 percent delinquency rate on the GSEs’ traditionally underwritten loans. While the vast majority of CRA loans were fixed rate and did not have higher interest rates like other subprime loans, Pinto’s research has found that most had high risk characteristics such as small downpayments and impaired credit, signaled by a low borrower FICO score. As a result, loans with these high-risk characteristics have performed similarly to fixed rate subprime loans.
Perhaps more important in its implications for the financial crisis, Fannie and Freddie did not report that their mortgages were subprime and Alt-A. So instead of the 12 million prime loans market participants expected, Fannie and Freddie had added 12 million subprime and Alt-A loans to the market...
So, by the middle of 2008, there were almost 27 million subprime and Alt-A loans in the U.S. financial system. This amounted to almost 50 percent of all mortgages. More than two-thirds of these mortgages were held or guaranteed by government agencies like FHA (about 4.8 million), and the GSEs Fannie and Freddie (12 million loans), and by U.S. banks (a residual of about 2.2 million).
But I personally think the real answer is, we just don't know, CRA lending wasn't tracked. Hopefully next time we decide to experiment on this scale, we collect some data. ;)
[1] http://www.clevelandfed.org/research/commentary/2000/1100.ht...
[2] http://www.federalreserve.gov/communitydev/craloansurvey.htm
[3] http://dallasfed.org/ca/bcp/2009/bcp0901.cfm
[4] http://www.aei.org/docLib/Pinto-Sizing-Total-Federal-Contrib...
[5] http://www.cato.org/pubs/journal/cj30n2/cj30n2-12.pdf