Spanish Bank Stress Tests
The Asset Quality Review and Bottom-Up Stress Test Exercise
I’m sure its fairly standard, but in the preamble, the report says it relies on 3rd party reports, that haven’t been verified or audited. Probably just standard legal mumbo jumbo, but worth mentioning since it is certainly lower than the standard a private equity investor would make.
The report covers the top 14 banking groups and about 90% of total domestic credit of the Spanish financial system. Personally I would assume the 10% not covered is a mess, but we can ignore that for now.
Conclusions
The report as a whole is devoid from reality. It is very accounting focused. While there is much to be commended by the apparent effort to analyze losses on a loan by loan level, the real world tends not to react that way.
I don’t think they fudged much in the loan by loan loss analysis, but any methodology that appears so detailed is as likely to have been designed to create loopholes as to have exposed them. They avoid assets outside of Spain and sovereign debt holdings. That is a major omission, especially because both add to the “absorption capacity”.
The tests seem to have been designed to create as much future income as possible for the banks, so as to limit the need for capital. The adverse scenario seems to be focused on 2012, and virtually all of the difference between adverse and base is this year, which seems ridiculous, and in my opinion is likely there to ensure minimum capital requirements. Maybe I’m wrong, but it just doesn’t seem to make sense to have most of the variation in the near term.
Lots of detail, some in depth work, but too accounting focused and strange to prevent a bank run. If we get another downturn, the banks will once again be targets, and this stress analysis won’t help much, because it isn’t designed to ensure extremely well capitalized banks. It is designed to sound bad while making the number as small as possible and ignoring too much to be truly useful.
There are some more comments on a section by section basis below and more to follow if I can get through more mind numbing pages.
Excluded Assets
The report doesn’t include foreign assets, fixed income and equity portfolios, and sovereign borrowing.
So foreign assets don’t count? I find that interesting. When you lend money to a bank, or better yet, inject capital, the foreign assets would seem meaningful. Maybe some of the bank groups are primarily domestic, but any true stress scenario should cover their foreign assets.
Ignoring the sovereign borrowing seems just absurd, at least in terms of any real stress. So as a starting point it is clear, the Spanish banking groups CANNOT withstand a sovereign default or PSI. No matter what the report says, this should be an important assumption that you use. Spanish banks cannot survive a sovereign default or significant PSI. That is conjecture on my part, but I think fits the facts, especially since Spanish banks, post LTRO, own an even more disproportionate amount of Spanish government debt.
I have no clue what the fixed income and equity portfolios are. I will assume those are trivial for now and fill in as I read more.
Stress Scenarios
A cumulative 6.5% drop in GDP for the 3 years from 2012 to 2014 doesn’t strike me as particularly onerous. It is stressful, but for many, matches a base case prediction rather than continued deterioration.
Unemployment peaking at 27.2% seems like we are close to a bottom since it is at 24.7% already and climbing. Since this stress test appears to be just for banks and ignores sovereign risk, it is probably fine that they ignore the budget stresses that continued high unemployment creates. But in the real world, hard to imagine investors don’t worry about the bank sovereign debt holdings.
The stress includes “additional drops” in house prices of 25% and land prices of 60%. I assume the land price drop is so severe because banks don’t own much raw land. I need to dig into details on house price, but many smart people expect a 50% peak to trough loss in house prices. If 25% is on top of 25% already taken, then maybe it is getting in the right ballpark. My suspicion is that Spanish banks haven’t marked down their housing related books that much, but will keep digging.
If you need a chuckle, you need to read which institutions are on the “Expert Coordination Committee” . Between them, I think they have managed about 0 dollars of money for profit between them all.
Loss Projection vs Loss Absorption Capacity
Two distinct, yet related concepts will play a key theme. The first is loss projection. That tends to get the initial focus. The amount of alleged absorption capacity is just as important, and may well turn out to be more interesting.
It looks like there was an attempt to be granular and detailed in the loan loss assessment. I’m suspicious but on surface looks like they tried.
On the loss absorption, they used business plans in part. If the Spanish banks had met their business plans in 2008, or 2009, or 2010, or 2011, we wouldn’t be in this problem. Using business plan projections strikes me as overly optimistic. Supposedly they try and adjust for this, but I find it hard to believe.
The mention a 3% CAGR deposit reduction under adverse scenarios. Without completely understanding that, it seems low relative to recent rates of deposit reduction and that is in an environment better than the stress case.
The banks benefit from “deleveraging” by estimating RWA reductions. Hmmm, I have yet to see much deleveraging, and it is incredibly hard to deleverage in a declining market, such as the adverse scenarios should have. More details to come, but seems unlikely to be correct, and if assets are defaulting or deteriorating during the scenario, the RWA would generally be going up.
€270 Billion for the “in-scope domestic back book of lending assets”
Say that 3 times fast. I remember at Bankers Trust that according to syndicate, we were number 1 in B HY bond issuance. Ranked by issues less than $500 million. For sponsor led transactions only. Whatever the in-scope domestic back book of lending assets is, I bet it excludes some other nasty stuff.
The losses are “only” €183 billion under the base scenario. So it is useful to know that the stress scenario added €90 billion. It is a bit concerning that the base scenario losses are so high, and as a percentage of assets, it doesn’t look like the stress scenario is that stressful relative to the base case.
€252 of system wide total loss absorption capacity
Why do we care about the level of absorption across the system? Maybe they get more granular later, but if a good bank is well reserved and a bad bank isn’t, the likelihood of the bad bank defaulting isn’t helped by the good bank’s reserves.
The report uses the existing provisions of €110 billion, “directly absorbing 40% of the total projected losses for the system in the adverse scenario”.
The domestic business will make €39 billion in profits during the full period. Definitely bank book treatment where anything that is “performing” generates income. History shows us that the markets won’t care about bank book treatment when analyzing the risk of an institution during times of stress. It will create its own estimate of unwind value and it won’t match anything like this. But that is the carry, and at least some of it seems realistic since they own lots of Spanish government debt, funded by LTRO, so what could go wrong?
Then, the banks make €22 billion in the Rest of the World. Yes, foreign assets pose no risk, but generate income.
Capital buffer generates approximately €73 billion of extra loss absorption capacity in the adverse scenario vs €22 billion in the base scenario. Say what? I need to check out this one, but more capacity to absorb losses in adverse scenario?
“Newly generated Deferred Tax Assets” have some limited use in loss absorption. This is growing ever further into an accounting gimmick exercise than a real world stress test.
Loss Absorption Goes Up in Adverse Scenario
I’m sure it will become clear to me, but in the base case if Figurer 2, the “system” has €183 billion of losses, with €212 billion of loss absorption.
While losses go up in the adverse scenario, to €270 billion, the absorption capacity jumps to €252. Asides from some tax savings, I’m still struggling with a real world view that banks can absorb more losses as losses increase.
Base vs Adverse
I am confused by a few things in Figure 6 which is the Macroeconomic scenarios provided by the Steering Committee.
Housing prices -5.6% in 2012 vs -19.9% in 2012. It is October now, don’t you think we would have a better guess than that? In 2013 the base case is -2.8% vs -4.5% and in 2014 it is -1.5% vs -2%. Something about this strikes me as very weird. The base and adverse converge and the time period we know the most about – the next 3 months, has the biggest divergence? Something smells here. I assume it is all about carry or default rates, or something, but it doesn’t strike me as right.
Then we get an interesting thing in the rates side. In the “adverse case” both Spanish debt and Euribor have higher rates than in the base case. Spanish yields I can understand, but my guess is that probably helps banks. They can own more bonds that have no stress risk and get more carry! Euribor, is trickier since it does seem to have some flight to quality risk. On rates, I’m not sure the impact, but if a bank is borrowing from LTRO it is not subject to Euribor, if it is lending to customers, it is not lending at Euribor, so it may well be that in the scenario analysis these higher rates help the banks as it shows more income.
Equities, which don’t count, are included. In 2014 the base case and adverse case are both 0%. In 2012 it is -1.3% vs -51.3%. Seriously, -1.3% vs -51.3%? Those numbers seem exact and random at the same time, in other words either irrelevant (since equities were explicitly excluded) or goalseeked.
Loan Losses
It looks like lots of efforts, including appraisals were used to determine realistic values. It looks like they had a very strict methodology for evaluating loan losses. It looks organized and thorough. So at least they have a methodology, but methodologies can be designed just as easily to create loopholes as to uncover them.
The “total in-scope domestic credit assets” (I really like that term) were about €1.4 trillion of performing and non-performing credit portfolio and a further €88 billion was already foreclosed on.
A far simpler methodology would have been to say, 50% loss on foreclosed, in both cases. That is a loss on foreclosed assets of €44 billion. Then let’s assume 10% loss on performing in base, and 16% in stress, giving rough totals of €180 and €270.
Yes, far more simplistic than what they used, but my gut is if the base case is remotely right, the stress case isn’t only that little bit worse. Yes, granularity can help, but in the end, values tend to correlate and everything gets dragged down, driven by the cheapness of the worst performing assets.
Almost funny, but a few pages lower, they show that almost all of the additional losses, via their methodology, come from the performing asset category. I will admit that by this point, I am starting to get lost in the details, in no small part, because I have a growing sense that the details don’t match what would happen in reality.
That timing to default was controlled in such a way to maximize “absorption capacity”.