On the Stability of Synthetic CDO Credit Ratings

• DOI: http://doi.org/10.1111/infi.12086
• Published date: 01 June 2016
• Author: Javier Zapata,Arturo Cifuentes
• Date: 01 June 2016

On the Stability of Synthetic

CDO Credit Ratings

Javier Zapata

y

and Arturo Cifuentes

z

y

Department of Statistics and Applied Probability, University of California,

and

z

Faculty of Economics and Business, Financial Regulation Center,

University of Chile.

Abstract

Synthetic collateralized debt obligations (CDOs) performed very badly

during the subprime crisis: they suffered massive rating downgrades

(even at the most senior levels of the capital structure) and inflicted

significant losses on investors. Using numerical simulations, this study

shows that such structures are highly unstable; minor errors in the

basic assumptions could manifest dramatically in the accuracy of CDO

rating calculations. Regardless of the quality of the underlying assets, it

is impossible to make reliable statements regarding the future perfor-

mance of a synthetic C DO tranche. Moreover, this st udy demonstrates

that single-point credit risk estimators (in which no attempt at specify-

ing a confidence interval is made) could be especially misleading.

Finally, the study suggests that a regulatory framework based on credit

ratings as they are presently defined is unlikely to be effective.

I. Introduction

The subprime crisis (2007–08) was a major disrup tion to the glo bal financial

markets. A salient feature of t his crisis was the fac t that most struct ured products

International Finance 19:2, 2016: pp. 201–218

DOI: 10.1111/infi.12086

© 2016 John Wiley & Sons Ltd

performed very p oorly, and their ratings reflec ted this reality. For example, most

collateralized deb t obligations (CDO s) included a tranche (t ypically th e top tranche)

that received an Aaa rating by at least t wo rating agencies. Neverth eless, according

to Benmelech and Dlugosz (2009), by 2009 more than one-third of such tranches

had been significantly downgrad ed, and many had d efaulted. Clear ly, such an

outcome was –at least conceptu ally speaking –unexpected.

It could be argued that the bad per formance of these tran sactions was simply due

to the poor quality of the unde rlying assets –name ly, subprime mortgages. In fact,

CDOs supported by other assets (e merging marke t sovereign debt, for instance)

weathered the subprime crisi s reasonably well. However, the question remains a s to

whether there was some thing peculiar a bout structured products th at made their

ratings perform so badly.

Leaving aside the fa ct that numerous investors are scept ical about the reliabili ty

of credit ratings, r atings are still a n essential component of th e financial regulator y

framework in most ma rkets. Moreover, the recent US legislation know n as the

Dodd–Frank Act gives paramou nt importance to rat ings, stating that one of its

objectives is ‘to promote accura cy in credit rat ings’(US Congress 2010). Moreover,

many financial insti tutions must comply w ith a number of ratings-d riven rules

concerning capital reserves and the type of ass ets they are allowed to h old.

There is well-supported conce rn amongst market par ticipants that the r atings of

structured products behave in an idiosyncr atic manner. Specifically, serious discrep-

ancies have been repor ted between the p erformance of corporate, municipal and

structured-products ratings ( Benmelech and Dlugosz 2009; Cornaggia et al. 2012).

Similarly, Hull and White (2010) reported inconsistencies between triple-B ratings

assigned to corporate bon ds and mezzanine tr anches of re-securi tizations, and

Gibson (2004) explored the behaviour of some simple synthetic CDOs and con-

cluded that their rati ngs were inconsistent with those of corpor ate bonds.

It is not surprising that, in this context, regulators an d government entities have

directed significant attention towards understandin g what, if anything , went wrong

with the ratings of st ructured products. What is sur prising, however, is that the

emphasis in these effor ts has almost exclusively bee n placed on the qualitative

aspects of the ratings industry –namely, conflicts of interest within the ratings

agencies, the p oor incentives of the c urrent fees and payment ar rangements and the

need for more competition in the c redit ratings market. Much less con sideration has

been given to the quantitative a spects of credit ratings. Questions such as wh ether

the ratings of CDOs are inherently unstable or whether the current ratings’metrics

capture the relevant risk features remain only partially explained.

Thus, it is clear th at a richer understan ding of the dynamics b ehind CDO rating s

is necessar y. From this perspective, this paper se eks to: (i) identif y the reasons for

the poor performance of such r atings during the subprim e crisis and (ii) exp lore the

implications of th e potential lack of stabilit y of such ratings in terms of financial

regulation.

202 Javier Zapata and Arturo Cifuentes

© 2016 John Wiley & Sons Ltd