Accounting for sovereign stress | FT Alphaville

Accounting for sovereign stress

Or the trouble “reviewing bank capital positions” — euphemism du jour from the European Banking Authority.

The EBA’s denied on Thursday that they’ll do actual new stress tests on banks in order to model bigger write-downs of peripheral sovereign debt and possibly identify where more capital is needed, as the FT reported. This worries us regarding how useful any EBA input will be.

Basically, it suggests that if the EBA are remodelling sovereign write-downs without a new stress test, they are going to apply the old stress tests’ cruddy rules. That means stuff like loss mitigation, provisioning, banks’ funding assumptions, and in particular an old chestnut round these parts — where sovereign debt gets classified in the accounting.

Market values vs recovery rates

Here’s an odd thing from the FT story:

The EBA, which is mid-way through a two-day crisis board meeting designed to assess the potential hit of mass sovereign restructurings, will use market values, to set “haircuts” on banks’ sovereign holdings.

Essentially FT Alphaville’s question is this — shouldn’t it be recovery values, not “market values”?

The old tests showed the impact on banks’ sovereign exposures in their trading books when distressed market values were fed in, but what about the rest?

(To remind, the EBA stress tests divided sovereign bond holdings into what they called the “trading book” and the “banking book”. In real world accounts, what matters are the classifications “held-for-trading”, “available-for-sale”, and “held-to-maturity”. Hold this thought for later.)

Ideally, you should be trying to test for permanent sovereign bond impairments. This raises the question of what guidance, if any, should be taken from “market” values at all. Especially when faced with the quagmire that is sovereign restructuring. Over to the accountants then, for some guidance on what actually counts as an impairment.

Bean-counting impairments

In the words of IAS 39 (via Deloitte):

A financial asset or group of assets is impaired, and impairment losses are recognised, only if there is objective evidence as a result of one or more events that occurred after the initial recognition of the asset. An entity is required to assess at each balance sheet date whether there is any objective evidence of impairment. If any such evidence exists, the entity is required to do a detailed impairment calculation to determine whether an impairment loss should be recognised. [IAS 39.58] The amount of the loss is measured as the difference between the asset’s carrying amount and the present value of estimated cash flows discounted at the financial asset’s original effective interest rate. [IAS 39.63]

Got it? Then onto some..

Case studies

The Greek swap, in its present incarnation, has featured banks marking to the profile of the impaired cash-flows set by the swap — a 21 per cent cut to net present value — because they think it’s the only permanent impairment event in prospect. This is therefore in line with definition IAS 39 above.

However, with the latest tranche of funding not yet released by the Troika, the swap may well be rewritten, resulting in further write-downs, or simply collapse altogether. The point being that it’s difficult, even when there are some cash flow details available, to objectively determine whether a impairment event has occurred and what the size of it is.

Going back to the market though, to see if there is any impairment guidance there, one runs into some very distressed Greek bonds. But the size of the discount on them isn’t the same as the recovery rate that would be offered by the sovereign upon default. Some banks don’t even believe there is a market any more. Of course, one of them is Dexia – and ironically some markets don’t even believe there’s a Dexia any more.

Leaving Greece and crossing the Ionian sea, Italian bond prices are under some pressure, but even contemplating a permanent impairment remains very tricky. And again, market values are just not any sensible guide to the recovery rate on defaulted Italian bonds.

You can see why it’s tricky for the EBA to wave around “market values” or the rules of the old stress tests to review capital for a scenario where Italy (or Spain etc.) restructures its debt.

Impairment, meet Core Tier 1

But there’s also a weird feedback loop at work here.

Admittedly the stress tests are just that – stressed scenarios. Stressing assets as if everything went horribly wrong is not the same as assets actually going wrong. But what is the distance between any potentially new stress test scenario and reality? Will the stress test’s parameters count as any form of “objective evidence” needed to declare an impairment? Or more to the point, what will auditors think?

If an asset is available-for-sale, and a lot of the sovereign bond holdings are, then mark-to-market losses do not travel via one’s profit and loss. Instead, such losses skip the income statement and go straight to equity without hitting Core Tier 1 capital. This is because Basel II doesn’t make banks account for MtM losses on AFS holdings when calculating CT1. And while Basel III won’t allow that, it’s not going to kick in any time soon.

So imagine that (say) a Italian bond held as AFS is trading at a 5 per cent discount, and the EBA now reviews capital positions assuming a 20 per cent impairment on that same bond. Will the auditor want that 5 per cent discount to at least be taken as a permanent impairment if up until that point it was actually just a MtM loss that didn’t hit CT1?

In other words, will the stress tests, that are meant to determine how much additional CT1 is needed, actually destroy CT1 on their path to righteously solvent banks?

Aim down, towards your foot… good, good, a little bit to the left… fire!

By Joseph Cotterill and Lisa Pollack

Related links:
Stress, mitigated – FT Alphaville
Eurozone bank stress test calculator – Reuters Breakingviews