In Part 1, FT Alphaville discussed the recent resurgence of so-called “BISTRO-type” securitisation deals. These allow banks to lower their capital requirements and defer losses by buying protection on portfolios of assets.
The Basel Committee issued a letter about such securitisation deals in December, effectively warning banks that over-engineering purely to reap regulatory capital benefits would not be tolerated. More recently, the head of the Financial Stability Board has called for the shadow banking sector, where the risk from such deals is offloaded, to be dragged into the harsh light of day.
Whether such deals actually get the treatment banks desire is for individual national regulators to decide, which is why we now wish to turn your attention to some guidance issued by the Financial Services Authority in the UK back in May. It seemingly flew under the radar at the time. The guidance came into force in September.
What the FSA guidance does, that the more recent rumblings don’t, is discuss the role that the internal models of banks have to play in getting these securitisations to work out favourably for banks. It also give some sense of the scale of the deals that had been done through 2010 — in the UK at least.
Model wars
As we discussed previously, these securitisations pool assets and then the bank buys protection via a credit default swap from an outside investor, typically a shadow bank like a hedge fund, on a slice of the deal.
Being allowed to lower capital requirements, on the back of such a deal, depends on being able to demonstrate to one’s regulators that some of the risk associated with the pooled assets has been transferred to an outside party.
To measure risk, and its transferredness, banks have a choice of models that they can use: standard models outlined by regulations that are formulaic, or internal-based ratings models which are custom-built by the bank.
According to the FSA, the custom-built models give some rather over-optimistic answers. Here it is in Regulatorese, from the guidance letter:
We have significant concerns that firms’ use of the SFM undermines the significant risk transfer requirement (SRT) with the reduction in RWEAs due to the use of the SFM being disproportionate to the credit risk transferred.
Dictionary:
SFM (Supervisory Formula Method) = custom-built model
RWEA (risk-weighted exposure amounts) = the amount of the pool of assets that the bank has to hold capital against
In summary, when the banks use their own models, answers ensure that are to the liking of the banks, and the regulators aren’t so happy about that.
The FSA has at least one thought as to why the custom-built models are doing such a bad job (emphasis ours):
The underlying formula contains an implicit assumption that there is no systematic risk in tranches of diversified portfolios that attach at a level of credit enhancement above the capital requirement on the underlying portfolio. However, the performance of senior tranches of many securitisations since 2007 has shown this assumption to be flawed.
In other words, the models don’t incorporate a systemic risk component to the extent that they should, given the lessons learned in the crisis.
In addition, where a firm’s IRB model proves, ex-post, to have under-estimated capital requirements on the underlying portfolio, the SFM leverages any undercapitalisation.
And the magnitude of that errors in the custom-built models can have a rather large impact, which ultimately means that banks come out with an answer that says they can hold even less regulatory capital.
As a result, the SFM will very often fail to appropriately capture the risks in retained securitisation positions.
That is, a lot of the risk is still actually sitting on the bank’s balance sheet, only now the bank isn’t actually holding enough capital against it in case of losses.
Ratings agencies to the rescue!
So what is the FSA doing about this?
Well, now it gets very Twilight Zone… they recommend that banks get these securitisations rated by an external agency, because external ratings can be plugged into the formulaic standard models, and those are presently giving out answers that regulators prefer the look of.
This at a time when US regulators are busy trying to figure out how to extract ratings from the rules.
Promoting the use of ratings by the likes of Moody’s and S&P is like using duct tape to fix shoes, it’s just not the right tool for the job. Anway, here it is in Regulatorese:
…the FSA will generally expect firms to obtain a public rating on retained tranches to apply the Ratings Based Approach (RBA) instead of using the SFM.
Sizing it up, in the UK at least
By now, you should be rightfully curious about how much of this stuff is out there. This brings us on to the most insightful paragraph in the FSA’s guidance, which we’re going to break down into bite-size chunks.
The number of firms reporting securitisation exposures in FSA004 fell from 30 to 23 from 31 December 2009 to 31 December 2010. 1
That fall isn’t what you think it is as here is the associated footnote:
1 The reduction in the number of firms reporting securitisation exposures was due to mergers, acquisitions and firm failure.
For those still standing though:
In the same period, the number of these firms reporting securitisation exposures under the SFM increased from three to six. This represents an increase from 10% to 26%.
Translated: more banks are trying to realise the benefits of using custom-built models.
The total securitisation exposures reported in FSA004 decreased from £272bn to £211bn from 31 December 2009 to 31 December 2010.2
Again, this fall isn’t necessarily what you think it is (or maybe it is…):
2 We attribute this decrease to firm failure, disposal of securitisation positions, transactions coming back on balance-sheet (and therefore exposures will be reported as credit risk exposures) and positions maturing.
Back onto the trend around custom-built models:
In the same period, the total exposures reported in FSA046 under the SFM increased from £5.6bn to £9.7bn. This represents an increase from 2.1% to 4.6%, this would increase to 7.5% if the transactions by UK banks currently under review by the FSA were permitted to use the SFM as proposed.
And in conclusion:
Therefore, the potential scale of firms’ use of the SFM for capital relief purposes is significant and the impact of any undercapitalisation due to the deficiencies in the supervisory formula could become systemic.
Too bad these FSA reports aren’t public. It’d sure be interesting to know how much this has grown through the end of 2011…
The best we can do is cross our fingers and hope they issue more guidance on the topic since they are required to have a “Materiality” section, which is where these kind of figures typically pop up. We’ll shoulder the goat-herding risk, bring us transparency!
Related links:
Back to the BISTRO for today’s securitisations, Part 1 – FT Alphaville
More Angry Birds, less regulatory arbitrage, please – FT Alphaville
Regulatory arbitrage: Basel is watching you – FT Alphaville
Risks emerging from shadows look worryingly like 2008 – FT
Basel 2.5 prompts flurry of asset sales and risk transfer deals – Risk
Gillian Tett on the Genesis of the debt disaster – FT (2009)