Friday, July 24, 2009

It’s Time for an Independent Audit of Rating Agency Corporate Bond Rating Performance

Posted on the RiskCenter by Donald R. van Deventer:

The U.S. Treasury has proposed new regulations on the major rating agencies, and a chorus of critics have attacked the agencies for inaccuracy of both structured products and corporate ratings over the last two years. One of my great fears has been the moral hazard that the rating agencies have to exaggerate the accuracy of their historical performance in rating corporate bonds. Today, those fears were realized. The U.S. Treasury should require an independent audit of rating agency performance in rating bonds and structured products. This post explains why.

The other day I received an e-mail from a former senior executive at Standard & Poor’s Corporation that included as an attachment an S&P study entitled “Annual 2008 Asia Corporate Default Study and Ratings Transitions, “ dated April 2009. A copy of the report is available here:

http://www2.standardandpoors.com/spf/pdf/fixedincome/Default_Study-Asian.pdf

One of the key credit modeling issues that my colleagues and I are dealing with currently is the parsing of government rescues into two groups of institutions, those who are deemed a “fail” for credit modeling purposes and those who are deemed a “not fail.” With this in mind, I turned to the S&P study for evidence that they too had taken the names that our client consensus has as “fails.” These include government rescues where the common shareholders have been disenfranchised and the government has effective control of the organization. Many names come to mind, but the clients we deal with are most animated when it comes to institutions like AIG, FNMA, and FHLMC. FNMA and FHLMC were rated AAA/Aaa even on September 7, 2008, the day that they were put into conservatorship by the U.S. government. AIG had been rated AAA less than four years before that firm too was effectively taken over by the U.S. government. Robert Jarrow and I argued in our March 16, 2009 blog post on http://www.riskcenter.com/story.php?id=17976 that the errors in corporate ratings were so severe that this era is essentially a “ratings Chernobyl” that would taint the accuracy of agency ratings for the next few decades.

With that in mind, I turned to the S&P study to see how serious the ratings accuracy damage was for the report through 2008. It turned out that Robert Jarrow and I were wrong—Standard & Poor’s just ignored its own errors so there was no “Ratings Chernobyl.” Table 15 in the S&P report is entitled “Comparison of Corporate Cumulative Average Default Rates” for both Asia and global corporates for the period 1981-2008. The one year default rate for global corporate rated AAA was ZERO! So was the two year cumulative default rate. The third year cumulative default rate was given as 0.09% and the fourth year cumulative default rate was 0.18%, so we guess that the third year reflects one default, the second year reflects 2 defaults and the implied number of AAA company years was 1/0.0018, or 556 “company years” of observations. What happened to FNMA and FHLMC? They’ve vanished from the data base! They didn’t fail!

“We didn’t default,” one FNMA friend of mine argued, “We paid interest and principal as scheduled.” Let me beg to differ. The common shareholders of both FNMA and FHLMC were totally disenfranchised and the government has assumed total control of both institutions to prevent their outright bankruptcy. The arbiter of the credit default swap market, ISDA, reports the following on www.isda.org with respect to FNMA and FHLMC:

“The 2008 Fannie Mae and Freddie Mac CDS Protocol (the “Protocol”) relates to settlement issues concerning Federal National Mortgage Association ("Fannie Mae") and Federal Home Loan Mortgage Corporation ("Freddie Mac"), two shareholder-owned companies mandated by the US Congress to provide funding to the U.S. housing market which were placed under a conservatorship by the U.S. Government on Sunday September 7.

“The purpose of the Protocol is to offer market participants an efficient way to address the settlement issues relating to credit derivative transactions referencing Fannie Mae and Freddie Mac. The Protocol will offer institutions the ability to amend their documentation for various credit derivatives transactions in order to utilize an auction scheduled for October 6, 2008 to determine the final price for certain Fannie Mae and Freddie Mac bonds. Markit Group Limited and Creditex Securities Corp. will administer the auction.

“The 2008 Fannie Mae and Freddie Mac CDS Protocol is open to ISDA members and non-members. The Protocol will be open between Monday September 29, 2008 and Thursday October 2, 2008.”

The two institutions failed and a formal auction of their securities was executed as is standard for failed firms under the normal terms of credit default swaps. For common shareholders, the “loss given default” was 100%. Jerome Fons pointed out to me that both firms also suspended dividend payments on preferred stock, as described in this entry from www.wikipedia.com.

http://en.wikipedia.org/wiki/Federal_takeover_of_Fannie_Mae_and_Freddie_Mac

For bond holders, because of the largess of the U.S. government, the “loss given default” turned out to be effectively zero, but the two firms still failed.

It’s impossible to know whether or not AIG is correctly reflected in the S&P table but it’s clear that FNMA and FHLMC are not. Similarly, what about Bear Stearns, Lehman Brothers, and the many other formerly highly rated firms that failed? Are they correctly reflected in the performance statistics? It’s impossible to know without an independent audit. The following chart assumes that all rating categories have a correct cumulative default rate except the AAA category. We raise the cumulative default rate by 0.18% for the AAA category to correctly reflect FNMA and FHLMC failures and compare that with what was originally reported.


Year 1

Year 2

Year 3

Year 4

Year 5

Year 6

Year 7

Year 8

Year 9

Year 10

AAA adjusted for FNMA and FHLMC

0.18

0.18

0.27

0.36

0.45

0.55

0.58

0.65

0.69

0.73

As originally reported











AAA

0.00

0.00

0.09

0.18

0.27

0.37

0.40

0.47

0.51

0.55

AA

0.03

0.08

0.14

0.25

0.34

0.45

0.56

0.65

0.73

0.83

A

0.08

0.20

0.34

0.52

0.72

0.95

1.21

1.45

1.69

1.94

BBB

0.24

0.68

1.17

1.79

2.43

3.06

3.59

4.12

4.63

5.16

The chart shows that the adjusted AAA cumulative default rate was higher than the AA cumulative default rates for years 1 to 7. The AAA default rate, after the adjustment, was higher than the A rated firms’ default rate in year 1 and almost equal in year 2. In year 1, the AAA default rate was 18 basis points, not much different from the 24 basis point loss rate from BBB rated companies.

The adjusted AAA default rates make the lack of accuracy in ratings performance starkly apparent. Why weren’t FNMA and FHLMC correctly reflected in these tables? Only an insider at S&P could answer the question. One might say “ratings are issue specific and reflect expected loss, not default, and therefore the fact that FNMA and FHLMC were excluded is not a mistake.” If that argument were valid, the table wouldn’t be described as listing corporate “default” rates. There wouldn’t have been an ISDA auction and a fail for both firms under the standard language of the credit default swap market.

What about Moody’s? A quick review of Moody’s self assessment shows that Moody’s has also omitted FNMA and FHLMC from their list of defaulters.

What is clear, however, is that rating agencies have a huge commercial incentive to use rose colored glasses in evaluating their own performance. It’s a conflict of interest of massive proportions. The U.S. Treasury should insist on independent third party audits of all rating agencies’ ratings performance statistics for anyone seeking Nationally Recognized Statistical Rating Organization status.

No comments: