The standard narrative surrounding the 2007–8 financial crisis focuses blame on the mortgage market and particularly on the supposedly faulty AAA rating of mortgage-backed securities. As that narrative goes, people knew these securities were junk and yet the securities traded in the market as if they were completely safe. When they turned out to be risky, the whole system came crumbling down.
A recent NBER working paper by Juan Ospina and Harald Uhlig performs a postmortem on the 2007–8 financial crisis and calls this narrative into question. Ospina and Uhlig have done the noble work of assembling data on all non-agency residential mortgage-backed securities issued up to 2008 and their performance up to 2014. The authors find seven main facts that they argue should collectively make us rethink the standard narrative.
1. The bulk of these securities were rated AAA.
Almost 87 percent of the principal amounts had the highest possible rating. This fits within the narrative that the AAA securities were an important part of the overall market.
2. AAA securities’ accumulated losses were 2.3 percent.
Contrary to the common narrative, the highest-rated securities did okay. Losses of 2.3 percent are historically large for AAA securities, but this is hardly the number the common narrative would imply. Other ratings did much worse (as we would expect), such as non-investment-grade bonds, which lost over 50 percent.
3. The subprime AAA-rated segment did well.
The standard narrative goes most astray with its focus on the subprime market, which had losses of 0.42 percent. Again, that is high for AAA securities, but it is not the level of destruction one would expect after reading popular books on the crisis.
4. Later vintages did worse than earlier vintages, except for subprime AAA securities.
Here the standard narrative gets things half right. Securities issued later did worse. For example, there were nearly no losses on AAA-rated securities issued before 2003, but losses were almost 5 percent for securities issued in the years 2006–8.
However, this timing is not true for subprime AAA securities, which receive the largest blame. Loss rates there were below 0.7 percent for the subprime AAA securities issued in the years 2006–8.
5. Losses were concentrated on a small share of securities.
Aggregate losses were low because about 65 percent of the securities lost nothing or less than 5 percent. However, nearly 20 percent of all securities lost more than 95 percent. That is significant and fits with the common narrative that some securities were awful.
6. The misrating for AAA securities was modest.
To see whether the ratings were systematically off, the authors compare ex ante expectations of returns given the ratings and ex post performance. Around 75 percent of AAA securities had little to no losses, which justifies their ratings. However, securities with ratings A and below had inflated ratings. Again, securities markets had problems, but the AAA securities were not one of them.
7. After accounting for the housing-price bust, the housing-price boom was good for the repayment of securities.
It was not the boom that was bad for repayment, but the bust. On average, an increase in house prices decreased losses, but the later decrease in house prices increased losses. This fact does not go against the standard narrative, but it is important to keep straight.
While we can debate whether we need to completely throw out the common narrative, Ospina and Uhlig have put facts on the table that any serious discussion of the financial crisis must take seriously.