Jul 01

How to think about securitization

by David Harper, CFA, FRM, CIPM

FRM |

For FRM candidates, the assigned reading on securitization is Culp's Chapter 16 from his outstanding book on structured finance. I like the entire book because it helps me to think holistically about structured finance. Given there are so many structural permutations, having a framework for the commonalities seems like a good way to handle the differences. From Culp, I takeaway the following key perspectives:

image

  • Motivation. The traditional motive is to remove loan assets from the balance sheet. Securitization is an alternative to the more obvious way to get assets off the balance sheet, to sell them. Within this motive, there are sub-motives running the gamut from perfectly natural to dubious (e.g., Basel I gave a regulatory incentive). The "balance sheet motive" starts with the originator. On the other hand, arbitrage is a motive that starts with the investor, or agents of the investor. An arbitrage CDO, for example, recruits a collateral manager who looks to build up the reference portfolio. Finally, the motive can be to simply monetize credit-sensitive assets.
  • Risk Transfer. The key feature of a securitization is the transfer of credit risk. In a true sale, the loan assets are transferred to a special purpose entity (or grantor trust). But, alternatively, the risk can be synthetically transferred: the originator can, instead of selling the loans, purchase credit default swaps (CDS). But it is not the same thing! There will be basis risk.
  • Reference Portfolio: the credit-sensitive assets can be physical debt (loans, bonds) or metaphysical (e.g., asset backed securities, CDO tranches)
  • Funded: is the credit risk funded (collateralized) or does it "merely" represent a promise by the counterparty. For example, the traditional balance sheet CDO is funded: the notes issued to investors are claims on cash flows (waterfall) where the assets are held by a trustee/custodian. But a CDS, expect for margin, is unfunded: there is no collateral but instead a promise.
  • Repackage risk into loss layers: This is a defining feature of structured finance: the risks and rewards that accompany credit-sensitive assets are sliced into a hierarchy of tranches. In a simple, traditional pass-through ABS, investors have equal footing and share proportionately in the cash flow. But tranched securitizations issue different debt types with different features to different investors.

So that a prototypical securitization consists of: Bank is motivated to monetize a reference portfolio credit-sensitive assets (while transferring credit risk and removing loans from the balance sheet). The cash is raised by investors who are issued securities that participate variously in the cash flow waterfall generated by the assets.

The subprime trust

We can apply this framework to the subprime case study (Ashcraft on the securitization of subprime):

image

  • Motivation. A traditional structure, no fancy footwork here! The originator (New Century Financial) sells the loans to a Goldman subsidiary, which in turn put them in a bankruptcy-remote trust. This is not a fancy, progressive structure. Rather, the reference portfolio happens to be subprime loans; put another way, given a few less defaults and much better pricing/modeling assumptions, the structure works fine.
  • Risk Transfer. Risk transferred to investors, indeed. But note the un-democratic and unexpected outcome: as of the report, the equity tranche had earned a +30.9% return! The excess spread was quite generous in the short term.
  • Reference Portfolio: mortgage loans
  • Funded: The structure is fully-funded. In fact, it is over-collateralized: the width of the equity tranche (1.3%) results from assets that exceed liabilities (i.e., debt securities issued to investors) by 1.3%.
  • Repackage risk into loss layers: as noted, unlike a simple pass-through, the GSAMP Trust is tranched into several classes: the equity, nine classes of mezzanine, and several senior classes

Comments

  1. Be the first to leave a comment!

Leave a Comment