In the context of discussions
during Basel II formulation, a lot less time and effort was spent on credit risk in
residential mortgages than on any other category. The fixed15%asset value correlation
is much less anchored on analysis of credit risk than almost any other part of the first
pillar of Basel II.
Suddenly in late 2006 and early 2007, all our attention turned to credit risk in
subprime mortgages, then to non-agency mortgages in general and to credit risk
in tranches of securitization of residential mortgages (residential mortgage-backed
securities (RMBSs) and collateralized debt obligations of RMBSs). Until then many
investors (especially those who had invested inAAAtranches) did not “look-through”
to the credit characteristics of the underlying pool of primitive assets. Look-through
has since become the norm and was included in the new amendments to the Basel II
rules in July 2009.
But there is a considerable difference between looking-through and analyzing the
cashflows of an RMBS on the one hand and “academic” credit risk analysis on the
other. This is because each securitization comes with a 100C page prospectus containing
numerous “nitty-gritty” tests and clauses that impact the cashflow to each
tranche contingent on some clause in the prospectus, some of which are reminiscent
of splitting hairs. Financial engineering has been squarely blamed for the lack of
transparency in, as well as investors’ poor understanding of, securitizations since the
beginning of the current financial crisis. Legal language in securitization prospectuses
is as much to blame for this lack of transparency though.
The prioritization of credit risk (a senior tranche gets paid first before a junior
tranche can be paid) so that investors can choose from a whole spectrum of credit risks
is at the core of the concept of securitization. The legal aspects are incidental, except
for the arm’s length nature of special purpose vehicles. The rest of the complications
that take up numerous pages of the prospectuses are not only dispensable conceptually but their removal is also essential to restore the securitization market to its former
glory. Life does not have to be this complicated!
Even today the most sophisticated players, including rating agencies, still seem to
be focused on the cashflows running through all the contingencies, which, in terms
of sheer numbers, far outweigh fundamental credit risk drivers such as probability of
default, loss given default and correlation. From this process, is it possible that they
may be missing the forest for the trees, not in the numerical answer, but in the fullest
understanding of what matters?
Something similar seems to have happened to many credit risk practitioners on
a broader scale when it comes to Altman’s observation. In our newfound attention
to credit risk in residential mortgages and RMBSs and in the furor over slackened
underwriting standards and poor documentation, etc, we seem to have lost sight of the
traditional signals of the corporate credit markets, which we all knew so much better
than credit risk in mortgages. In his paper, Altman goes on to argue that “what happened
in the corporate market was the catalyst for flight to quality and the subsequent
meltdown, not only in the credit markets but also in the real economy” and he then
discusses what to expect in terms of corporate defaults, bankruptcies and recoveries
in the near future.
In this issue we present two full-length research papers and two technical reports.
The first paper, “Modeling credit exposure for collateralized counterparties”, is by
Pykhtin and it pertains to the very topical subject of counterparty credit risk in the
presence of collateral agreement. The author proposes an interesting semi-analytical
method for calculating exposure at the time of a margin call without full Monte Carlo
simulation that makes calculations almost twice as fast. The emphasis of this paper is
on calculating “expected exposure” profiles. Expected exposure is a very important
quantity for several applications. Banks need to know expected exposure levels in
order to be able to price and hedge counterparty credit risk via credit valuation adjustment
calculation. The expected exposure is often used as a basis for loan-equivalent
exposure in economic capital calculations. Furthermore, expected exposure is the
required input for calculating exposure at default for the minimum capital requirements
under Basel II.
In the second paper, “Granularity in a qualitative factor model”, by Gourieroux
and Monfort, the authors provide a unified setting for factor models that are applied
in a number of areas in credit risk modeling: from the single factor correlation model
and its multifactor extensions to rating dynamics, all with panel data. These models
play important roles in credit risk measurement and management practices. Under
this unified setting, the paper examines the behavior of the estimators of these models
when the cross-sectional dimension is large and provides “granularity adjustments”
for the maximum-likelihood estimators of the factor sensitivities. Furthermore, the authors conducted a Monte Carlo study of the finite sample properties. The results in
the paper are relevant to researchers who study or use these models, especially those
who need to estimate the factor sensitivity parameters for applications in credit risk
measurement.
The last two papers in this issue are technical reports. A technical report describes
a particular practical technique and enumerates situations in which it works well
and others in which it does not. Such reports provide extremely useful information to
practitioners in terms of saved time and minimizing duplication of effort. The contents
of technical reports complement the rigorous conceptual and model developments
presented in the research papers and provide a lot of value to practitioners.
The first technical report, “Selecting credit portfolios for collateralized loan obligation
transactions: a heuristic algorithm”, is by Westerfeld and Weber. The paper
proposes a heuristic algorithm for selecting loans for a pool backing collateralized
loan obligations (CLO). The selection tries to maximize the unexpected loss of the
loans in the pool under constraints ensuring that the resulting CLO achieves some
required rating. In the structuring process of aCLOsuch an algorithm could strengthen
the negotiating power of the originator toward rating agencies and investors. Furthermore,
CLO transactions usually require a frequent reshuffling of the underlying credit
portfolio. This approach may be feasible and less time consuming for such applications.
The second technical report, “An improved multivariate Markov chain model for
credit risk”, is by Ching, Siu, Li, Jiang, Li and Li. The technical report presents a
multivariate Markov chain model of transitions of ratings of credit risky entities.
The implementation and the framework itself attempt, as a primary objective, to be
“parsimonious” in terms of the number of parameters. Calibration of the model to
historical rating migration data can be formulated as a linear programming problem
and it can generally be implemented using spreadsheets.
REFERENCES
Altman, E. I. (2009). Corporate credit: defaults, recoveries, and ramifications. CFA Institute
Conference Proceedings Quarterly 26(3), 47–54. |
