How Do You Rate?
In January, the US Securities and Exchange Commission (SEC) announced that Standard & Poor’s Ratings Services (S&P) had agreed to pay almost $80 million to the SEC and other regulatory agencies for a series of federal securities law violations involving “fraudulent misconduct in its ratings of certain commercial mortgage-backed securities (CMBS)” . S&P was also banned for one year from issuing ratings in the commercial bond market.
The SEC press release was scathing in that S&P “elevated its own financial interests above investors by loosening its rating criteria to obtain business and then obscuring these changes from investors”. A clear case of a ‘conflict of interest’ being resolved in favour of the firm rather than its clients. What makes this failure to manage conflicts of interest more appalling is that the events referred to occurred in 2011, well after the Global Financial Crisis (GFC) in which the main Credit Rating Organizations (CROs), such as S&P and Moody’s, were soundly criticised for not handling similar conflicts of interest.
In a separate order, the SEC began proceedings against Barbara Duka, who was an S&P managing director with responsibility for new issue and surveillance ratings of CMBS between 2009 and 2012. Ms Duka is contesting the accusations but it appears that she may be the ‘scapegoat’ upon whom all of the opprobrium will fall.
How are securities rated?
Providing a credit rating for any security is complex and is based upon assumptions about the future behaviour of the security, such as its probability of default, which in turn are partially based on historical behaviour of similar securities. The goal for any issuer of a security is to have it rated ‘triple A’ or AAA, that is highly unlikely to default. Critical to the rating that is provided by agencies, such as S&P, are the assumptions that are made about likely performance and the complex formulae used to arrive at the final rate. Whether these assumptions are defensible or not, the important point is that the assumptions must be disclosed to the sophisticated investors who purchase complex securities such as CMBS – caveat emptor!
The complex models used to rate securities are sensitive to the assumptions made, for example in this case a number known as the ‘loan constant’ involving the assumed rate of loan repayments, to which the final rating was extremely sensitive. And the loan constant was not the only key assumption, another constant, the so-called Loss Severity (LS), also played a part in the rating formulae.
To cut a long and convoluted story short. In its published documents, and in a critique of an issue rated by other agencies, S&P promoted the use of a conservative ‘stressed’ loan constant. However, it turned out that the use of conservative loan and severity constants resulted in lower ratings which effectively froze S&P out of the market, in effect they were losing business because their ratings were too low. To cut to the chase, S&P were found to have (slightly but significantly) relaxed the criteria for calculating the constants, resulting in higher ratings and more business. The crux of the SEC indictment was not that the criteria were relaxed but that investors were not kept fully informed.
How expert are experts?
It is not obvious however what the various assumptions should be – it depends on one’s point of view! And S&P hired some very talented and experienced people to take a viewpoint on the assumptions that formed part of their rating criteria. But there is nothing absolute about such criteria, one must decide what data is included in the calculations and more importantly what data is NOT included. These are value judgements by experts. But experts are just as susceptible to decision-making biases as non-experts.
Experts are naturally subject to ‘confirmation biases’, unconsciously searching for information that fits with their ‘expert view’ of the world. Experts are subject to ‘overconfidence’ in their predictions, since they are experts after all and are expected to be right in their own field. Experts tend to be ‘satisficers’, finding a solution that works in the sure knowledge that they can explain any discrepancies, the so-called ‘illusion of skill’. And experts are subject to all-pervading ‘anchoring bias’, in the case of S&P everyone was fixed on the magical AAA rating, nothing else would do.
Just because people are experts, doesn’t mean that they necessarily make good decisions!
When things go wrong there appears to be a deep-seated human need to find someone to put the blame on. The concept of a scapegoat or whipping boy is ancient, appearing in the Old Testament and pre-Christian Greek law and is well-established in human consciousness as providing a way in which our sins can be diverted to someone/something else who will take any punishment due. By assigning fault to someone else we can somehow convince ourselves that, with the scapegoat punished, the event will not happen again. This is a sort of confirmation bias convincing ourselves that, unlike the scapegoat, we are beyond reproach. But scapegoating does mean that we don’t learn from our mistakes.
Barbara Duka appears to have been assigned the role of scapegoat in the S&P settlement, mainly because she resigned before the events came to light. There is no doubt that Ms Duka should, as senior manager, bear some responsibility but the evidence provided shows that Ms Duka did not make decisions capriciously but as a result of discussions with subordinates, colleagues and her managers. The damming evidence against Ms Duka appears to be a he said/she said exchange between her and a senior manager where a change in criteria was discussed and Ms Duka took action to change the criteria but did not document the changes.
Sales picked up (because the criteria had been loosened) but no-one appeared to have demurred, especially as the changes were obvious to all analysts involved in the ratings process. No senior manager seemed to have questioned and investigated the reasons why the business had turned around, until some investors questioned the ratings that had been given. The disclosure to investors was changed but was not explicit as to actual criteria used. One of the S&P analysts stated in testimony that a new sentence in the disclosure had been “written to be vague . . . based upon her instruction.” So Ms Duka was solely to blame – I am only following orders! Ms Duka has been thrown to the lions.
Look in the mirror! How many of us can honestly say that we have never used weasel words to cover our lack of knowledge or to help justify a certain decision. We justify such obfuscation by claiming that the reader is sophisticated enough, or well-paid enough, to read between the lines and see what was meant. We know that being too specific sometimes will raise new questions that we don’t have the answers to, and if we have the answers the listener may not want to hear them.
When we write a report, or an academic paper, we have an audience in mind and we think we understand what that audience wants, and not surprisingly we try to give it to them. It takes a brave man or woman to give bad news to their superiors- it often is career limiting decision. Each firm’s style of report-writing is another but more subtle form of Groupthink.
The SEC accused Ms Duka of using ‘vague language’ and deliberately leaving out a reference to a particular assumption in the published disclosures. S&P staff knew of this and some even questioned it but were reassured by staff other than Ms Duka. Yes, Ms Duka bears responsibility for the wording of the final disclosure, but she was not a dictator, others concurred at least until the disclosure was questioned, then turned around to cast the first stone. In this case, Ms Duka may be like Alice in Wonderland in that she meant what she said about ratings, but didn’t quite say what she meant.
Scapegoating is a form of Groupthink , specifically excluding an individual from the group and in an Orwellian twist, removing his/her positive contributions from the narrative. But the change in rating criteria was known to many people in S&P and the SEC inquiry does not report any major objections. It does, however, report events that might indicate Groupthink.
In the main indictment, the SEC records that, in June 2012, S&P published details of its proposed new ratings criteria for CMBS and invited feedback and questions from market participants. The assumptions underlying the criteria were based on data collected during the Great Depression, some 80 years previously, and one of the Senior Criteria Officers (SCOs) responsible questioned how such assumptions could possibly be justified. After significant internal discussion, the unidentified SCO responsible added a new assumption which (magically) justified the new criteria but was dismayed when the overall group removed key information from the publication, noting, presciently, that one day he may be “sitting in front of the SEC”. Nonetheless, after the criteria was published, the SCO, in a self-assessment, congratulated himself, “In my role, I recognize the need to balance between the best theoretical solution and the best business solution.”
This is classic Groupthink behaviour where a group puts pressure on its members to conform and in conforming initial doubters come to self-justify and then actively support their changes of heart .
People Risk Management
The risk that an employee or group make decisions as a result of Groupthink is a People Risk as it can lead to significant losses even the bankruptcy of the firm. In this case a decision to finesse a disclosure resulted in an $80 million fine and considerable reputation damage. This blog is one of a planned series that will discuss facets of People Risk in general and Decision-Making in particular. It is obvious that managers and assurance functions, such as Risk Management, Audit, Compliance and Human Resources, must understand the concept of People Risk, particularly the influence of individual and group biases on decision-making, because badly-made decisions may result in significant damage to the firm.
 See “SEC Announces Charges Against Standard & Poor’s for Fraudulent Ratings Misconduct” http://www.sec.gov/news/pressrelease/2015-10.html#.VMBp0i4Ylb2
 Janis I., 1971, ‘Groupthink: The desperate drive for consensus at any cost’, in The Classics of Organisational Theory, (eds. Shafritz J. M. and Ott J. S.) Wadsworth
 See for example McConnell P. J. (2013) ‘Northern Rock – The Group That Thinks Together, Sinks Together’ Journal of Risk and Governance (2013) Vol. 2/2
Errata Note an earlier version inadvertently stated that ‘SEC’ rather than ‘S&P’ staff ‘knew of this and some even questioned it but were reassured by staff other than Ms Duka’.