Comment on Antoniades and Tarashev

Full note available here.

William Perraudin

An interesting recent article in the BIS Quarterly Bulletin by Antoniades and Tarashev argues that capital for mezzanine tranches of securitisations should be boosted substantially because of uncertainty about pool default probabilities. The authors claim that this is true even when the securitisations are Simple, Standard and Transparent (SST) in the terminology of the recent EBA discussion paper.

The authors perform an exercise in which they calculate what they refer to as the expected under-capitalisation for securitisation tranches of different seniority. They claim that under-capitalisation amounts to 60% for tranches with attachment points close to pool capital.

In this note, we point out that their strong conclusions are highly sensitive to the capital model employed. We argue that the one-factor model they emphasise is not appropriate for understanding securitisation risk. In its latest calibration of securitisation capital, the Basel Committee has adopted, instead, a two factor model. When this is employed (based on the author’s calculations) over-capitalisation drops to 6-7%. With the higher intra-pool correlation parameter we have employed in our work, the over-capitalisation would be still lower.

Last, the tranches on which the authors focus attach much lower on the capital structure than the bulk of what the European market regards as typical mezzanine tranches. (Mezzanine attachment points in Europe are generally around 2 to 3 times pool capital.) Hence, even taken at face value, their propositions are not directly relevant for most real-life European mezzanine tranches.