Updated (end of post)
The following post is attempting to put some numbers behind a highly uncertain, opaque and completely under-reported side of the Irish banks operations - the side relating to securitisations and derivatives exposures. My numbers below are pure estimates and their objective is to at least start raising the questions as to the depth of our (taxpayers) exposure to this murky world of banks' securitised assets.
Before we begin, I must also relay my thanks to Brian Lucey and 3 anonymous experts for providing advice and comments on the earlier draft and to LorcanRK who was involved in trying to scope the problem earlier.
Years ago, before our sick puppies (banks) became sick, in the golden days when the Anglopup, AIBickey, permo, INBiSquit, EBSsie and BofIpooch were still wagging their happy tails around the streets of Dublin, securitisation was all the rage.
The basic idea behind this transaction runs innocuously enough as follows: a bank holds a bunch of loans, say mortgages. These yield an annual revenue stream, but hold up capital, restricting new lending. To help unlock this capital, a bank can package these loans together and sell them to an SPV which will issue a paper security against these loans that entitles the owner to a share of the total package of loans as they yield returns over time. An SPV, of course, doesn’t manage the mortgages but leaves them in the custody of the bank which acts as a manager/custodian, responsible for collecting the moneys due and paying out to the SPV.
Now, for a bit relevant to us: an agreement between the SPV and the custodian has two key covenants:
- loans are held by the custodian in trust, so that the custodian is obliged, upon either the termination of the management contract or should other covenants be breached, to deliver the actual loans/mortgages to the SPV owner;
- ability of the custodian/manager to hold on to the loans is subject to a minimum credit rating, usually - investment grade.
The first point means that should an SPV ask an Irish bank for its loans (due to a breach in its covenants), the banks must deliver these loans.
The second point means that if the covenants are breached, by, say for the sake of argument, Irish banks rating sinking to junk, the banks can be found in a breach of covenants and face:
- a margin call – according to my sources, of up to a whooping 20% face value of the securitized loans in some cases; and/or
- a call on the actual loans to be transferred to a different manager/custodian nominated by the SPV
Every securitized contract runs alongside it a derivative security designed to protect against the risk exposures relating to the loans.
These derivatives can be
- symmetric – covering both sides of the potential exposure – e.g. interest rates swaps going both ways or
- asymmetric or uni-directional, covering only one side of the risk exposure (e.g. an interest rate swap insuring against a future rise in the interest rates).
The derivatives can be written by an independent entity or by the bank, but for the reasons of good risk management (maturity mismatch risk and direct exposure to underwriter risk) these derivatives should really be underwritten by the third parties, not the custodians.
Now, let’s go back to the history. Earlier this year, I wrote about our ‘national derivatives accounts’:
- AIB held the total derivative exposure to the notional value of €261bn in 2008 which fell to €197bn in 2009 (here)
- BOI held €360.5bn (here) in 2009
- Anglo held some €268.3bn worth of notional value derivatives in 2008 (here), falling to €184.5bn in 2010 (here)
The above is very close to the gross notional exposure amounts of €640 billion (for two banks ex-Anglo) reported in 2008 by the employee of the Financial Regulator - Grellan O'Kelly (here).
So now, suppose that the notional value reflects symmetric hedges, and even there, let's assume that directionality is such that benign risk is weighted by twice the weight assigned to maximum loss-linked risk, so that the underlying value of these derivatives is around 1/3rd of the €742.3 billion of notional value, or €245 billion.
Here is the beefy problem. Since these derivatives are written against real loans contracts, what happens if the covenants of the SPVs behind them are breached?
Let’s talk some hypotheticals (since we have no actual clarity on these):
- Scenario 1: Irish Government debt sinks to junk, which automatically means banks debt sinks to junk (while I was writing this, the latest Moody’s downgrade pushed it even deeper...). There’s a margin call on derivatives of say ½ of 20% mentioned above, or 10%. Oops – Irish banks are in a hole for up to 24.5bn off the starting line (10% of the 245bn above)
- Scenario 2: Instead of a call on the derivatives, SPV breaks management agreement with an Irish bank and asks for its loans to be moved out of the bank. Wouldn't be a problem, unless: what if the bank, in the mean time, has leveraged the same loans it held in custody for the SPV at the ECB (or CBofI or both) discount window? Well, should the SPVs insist, the Irish banks will be forced to buy their collateral out of ECB and CB of Ireland to the amount that the banks borrowed against such collateral.
Things are starting to smell rotten… But do not be afraid, those in charge who still have some brains left spotted the dodgy stuff. To our chagrin, however, the smart ones are in Frankfurt, not in Dublin. Back in August 2008, the ECB has pulled the plug on taking Irish banks-securitised loans as collateral. Miraculously, in the end of 2008, CBofI lent Anglo €10.5bn against some mysterious collateral that, several of my sources argued, was previously rejected by the ECB.
Why would the ECB decline to take securitised packages as collateral, while taking the loans? Surely this signals something is amiss with the vehicle of securitisation as carried out by the Irish banks?
Two things can be dodgy with the securitized packages in general:
- Underlying derivatives, and/or
- Security over the loans/assets that are securitized.
I am not going to speculate what it is – time will tell. Instead, let’s run through some scenarios on potential losses due to the above positions.
Assumptions:
- Assume that the above gross notional amounts of derivatives are 2/3 covering one side of exposure (e.g. expected increases in interest rates, for interest rate swaps) and 1/3 covering less expected opposite direction risk. This means that of the total values of derivatives written by the 3 banks, these derivatives were covering a collateralised pool of loans/assets equal to 1/3 of the gross notional derivatives.
- Now, some of collateralised assets were held by the banks themselves, but we do not know how much. So let’s assume that 25% and 50% are reasonable amounts for these shares, implying that banks sold on some 50% to 75% of the securitised assets
- Next suppose that the banks have written down these securitised assets by 20% (a gross overestimate, but let’s allow it to be conservative) and that the ECB has applied the usual 15% haircut in lending against the above writedowns
- Table below shows the estimates of potential losses
So the downside from the derivatives exposure and securitization can range between €12.25bn and €50.8bn.
Pretty wide.
Let’s take a look at the underlying assumptions. Running through the ‘What if covenants are breached?’ scenarios, one has to remember that many of the securitized loans borrowed against are related to more stable, longer-term mortgages. Since default rates across mortgages are lower it is highly unlikely that SPVs wouldn’t want to claim them out of the hands of the insolvent banks. This means that the 10% margin call on all loans scenario is highly unlikely to materialize. More likely – either the margin calls will be larger, or full call backs will be triggered. Which suggests that the range above more realistically should be expected around €17.15bn and €25.7bn.
Also, recall that Irish banks weren’t really at the races in speculating on financial instruments, preferring instead to speculate on property. This means that my assumption of 50% unidirectional net derivatives relating to property securitization is pretty conservative.
And remember that none of this has been factored by either the IMF or anyone else into the expected losses across the Irish banks. It hasn’t been incorporated into my earlier estimates of
- €67-70 billion total losses on NAMA, recognized losses and post-2010 commercial and investment books’ losses, and
- €9-11 billion total losses on mortgages post-2010, plus
- the lower €17bn figure as an estimate for the derivatives and securitization-related losses.
The total expected loss across the entire banking sector, net of recoveries might be as high as €93-98 billion. Or it might go as high as €107bn. And at this point, folks, even an old hawk like myself starts to feel scared.
Note: these are potential estimates. Given that we have been given no clarity as to the depth of securitisations, or the derivative instruments underlying it, nor do we have any idea as to what the banks have been doing with custodial-managed loans that relate to securitised products, one can only guesstimate - or speculate - as to the true extent of losses. I tried my best to be very, very conservative in the above, with my upper limit of factored estimate of €25.7bn in losses being below the average of the most benign scenario (€12.25bn) and the worst case scenario (€50.8bn). I was also very conservative in my assumptions. Note also that in the end, €17-25bn range of losses used in final estimate of the total cost of banks bailouts corresponds to just 2.29-3.37% of the notional value of all derivatives held in 2009 by the three banks.
Update: things are hardly trivial when it comes to potential securitisation-linked derivatives exposure. Back in 2007, the IMF has warned Irish regulators that:
“
BoI has transferred the bulk of its domestic residential mortgage assets to a designated mortgage credit institution, which has a banking license to issue mortgage covered securities.—these are used both for hedging interest risk and for generating additional funding. Almost 60 percent of these securities were held by other Euro Area members, while 25 percent was held in USD by other countries. (IMF WP/07/44: External Linkages and Contagion Risk in Irish Banks, by Elena Duggar and Srobona Mitra - here) Did IMF say 'the bulk'? So as of 2006-2007, the bulk of mortgages were out to securitisation in a 'conservatively' run BofI?
So while Manufacturing is signaling weak growth across both output and employment, Services are showing neither:
Weighted (by economy weights) average of the two points in the chart above places December squarely into the Recession Area along the axis that barely enters Optimal Growth Area. It is worth noting that longer-term trends (and these are strong with 0.847 RSq for Services and even stronger 0.892 for Manufacturing) do not support Jobless Recovery. In contrast with historical experience, this is exactly where we are heading in Q1-Q2 2011 per chart above.
