Trying to assess the Cajas' mess

We believe one single number can summarize the current mess the Cajas are in: 33,480. That’s the monthly sales of new homes in Spain at the peak the real estate and credit bubbles, in January 2007. In the meantime, France built roughly 180,000 homes a year and Germany 200,000.

Spain built more new homes than France and Germany combined, with only one third of their combined population… Ireland was a mess, too, with 93,000 new homes in 2006 and a population of only 4.5m.

Germany Ireland France Spain
New homes built (k) 250 93 180 400
Population (M) 82 4,2 61 44
Hab per new homes 328 45 339 110

NB: real estate statistics are notoriously difficult to handle due to various methodologies used by various sources. These numbers give an order of magnitude more than anything else.

So the real estate bubble was huge and now Spanish banks are in a mess. Two questions come to mind:

a) why does this comes into the spotlight so late, more than two years after the Subprime crisis in the US and

b) what is the exact size of the mess? We will also try to estimate the possible impact of a full stress scenario on financial markets.

1. Why is it so complicated to assess the potential writedowns of Spanish cajas?

We believe there are two main reasons for that.

The first one is regulatory. The Spanish Central Bank, unlike other European central banks, has always been very reluctant to allow Spanish banks to book RMBS/CMBS (wrapped or not) in their trading books and have forced them to put these in their banking books. As a consequence, the regulatory arbitrage that all other European banks were using (the negative basis trade with low margins, supposedly low risk - AAA tranches reinsured by AAA monolines – and very low capital charges, hence high ROEs) was not available to Spanish banks. This protected Spanish banks from subprime and other toxic assets and forced them to turn to their domestic market, with real estate financing booked in their banking books.

The good thing about the banking book is that you don’t book it on a mark to market basis and this allows banks to handle non performing loans smoothly over time. Spanish banks didn’t book the impressive write-downs that other banks had to book due to the strong drop of market prices of RMBS/CMBS paper. The bad thing is that it is less transparent than mark to market accounting and, eventually, you have to book the losses, only later.

This effect has been exaggerated by another Spanish regulatory “trick”: the Spanish regulator asks banks to use what is called “ex-ante provisioning” or “provision buffers” or “dynamic provisioning”, fancy names for provisions that are booked to cover the expected loss of a real estate loan portfolio before any actual loss. Over time, Spanish banks build up provisions that they are using when losses actually start to rise. It is only when their historical “buffer” is fully depleted that they start to show high number of losses and deplete their equity. This is why accounting figures of Spanish banks have been resilient during the first phase of the crisis. On top of that, the central bank also imposes rules on the time banks have to wait before they can impair a loan.

Analysts estimate that for listed Spanish banks these reserves will be fully depleted in one to three quarters, depending on the bank and other various assumptions.
Overall, the highly regulatory approach to the provisioning of real estate losses in Spain makes the system more resilient and less cyclical, but obviously less transparent and harder to assess precisely.

This is even more the case with the Cajas, of course, as that they are (mostly) not listed, and that is the second reason. The global Spanish sector is split roughly equally between the large listed banks and the cajas, and there is little information available on these, apart from some credit analysis from the rating agencies.

2. So let’s try to put a number …

Where can we find some relevant indication to assess the cumulated losses that the Cajas will eventually face and the ultimate capital needs of the Cajas ? Let us first look at the macro numbers of the sector and what has been going on for a year.

a. Current situation

45 Cajas in 2009, only 17 left in 2011... No bankruptcy here, but mergers sponsored by the FROB, the fund investing on behalf of the Spanish governement. FROB invested 11.6bn so far, mainly in the form of hybrid convertible capital (8 investments, each time roughly 1.4% of total assets of the Caja).

The Cajas sector can now be summarized as follows (source BOFA ML):
-1290bn€ of total assets, of which :
-217bn€ of exposure to real estate, of which :
-98bn€ of troubled exposure (NPL 29bn€, substandard loans 27bn€ and seized property 41bn€). This exposure is currently covered by :
-59bn€ of equity,
-30bn€ of provisions,
-and 94bn€ of collateral value, as estimated by the Bank of Spain (discretionary haircuts depending on the type of real estate).

Of course, these are aggregate data, and some Cajas are more in trouble than others, well capitalized entities : Unicaja has close to 13% Core Tier 1 and less than 8% of assets as troubled exposure, whereas Catalunya Caixa has above 15% of troubled exposures and just slightly more than 6% core tier 1... This is even more important as 3 Cajas represent more than 50% of the troubled real estate exposure (BFA, Banco Base and La Caixa).

Clearly, current provisions are underestimating losses in a stress scenario and it is now time to estimate the potential shortfall of equity for the global Cajas sector.

b. Historical comparison

History is a first guide. According to Nomura the Non Performing Loan (NPL) statistics are very close to the ones observed during the 1993 crisis, but the provisioning has been substantially different.

A very enlightning chart indeed, and the message is pretty clear: provisions are close to 1%, where they should peak at a maximum of 3%, or a 3 to 1 ratio.

Provisions at the Cajas are currently at 30bn€. Even though they represent more than 1% of total assets (closer to 2.3%) one can estimate that the 3:1 ratio would still apply (the higher provisions are simply due to more risks, not to different accounting methodologies). This would mean that an extra 60bn of provisions would be required.

However this does not immediately translate into a capital shortfall : some banks are adequately capitalized even if they will suffer high losses and all the Cajas will also be making money from profitable businesses.

However, Cajas’ core businesses not higly profitable and the Core Tier 1 will be increased due to Basel III and new Spanish capital adequacy rules. This is why we will keep the 60bn€ number, even though 75% of that number would be more realistic, using a back-of-the enveloppe calculation and CEBS methodology.

Note that rating agencies and analysts concur that large listed banks will be able to cover for these provisions with their own profit generation capabilities, and maybe some capital raising due to Basel III, so that approach wouldn’t apply to these banks.

c. Rating agencies analysis

We consider three different approaches :

1. Moody’s built a stress scenario for Spanish banks. The outcome, in case of continued deterioration of the Spanish economy, is cumulated losses of 225bn€, i.e. 175bn€ over current provisioning levels. This would lead to roughly 100bn€ required for the Cajas, but probably over a longer period of time (2-3 years).

2. Standard and Poors have built a rather impressive comprehensive model to build their own stress tests "a la CEBS" with a target Core Tier 1 at 7% (way above the current the regulatory minimum but still a bit low in our view to restore confidence in the banking sector). The final estimate of S&P is only 34bn€ for the cajas. Obviously, using a 9% Core Tier 1 would raise that number, but it would still remain below Moodys’ estimates.

3. BOFA Merril Lynch has recently published their own stress tests for the Cajas. Their methodology (including 20bn€ losses on sovereign risk) lead to 111bn€ of losses for the Cajas and translates into capital needs of 43bn€, assuming a Core Tier 1 at 8.5%. This is close to S&P estimates, taking into account the different Core Tier 1 ratios.

d. Comparing with other countries

Two relevant international examples come to mind: Nevada and Ireland. Actually both cases look the same and that is why Blackrock, acting as an advisor to the Irish Central Bank, decided to use the real estate market in Nevada, incredibly depressed, to perform the stress tests for Irish banks. Looking at Blackrock’s estimates for Ireland will actually give us a view combining Ireland and Nevada, sort of…

It is of course extremely difficult to compare real estate markets and actual loan losses in Spain and Ireland, but we propose to use two very simple rules of thumbs. We believe the situation is two to three times worse in Ireland than in Spain. Two, because the Irish built (roughly) twice more new homes than the Spanish and three because the size of the Irish banks’ balance sheet was (roughly) three times the size of Spanish banks’ balance sheet in terms of GDP%. Agreed, this might appear very simplistic but we believe it gives a reasonable order of magnitude of the problem.

Since we want to be highly conservative, we will use “pure” Blackrock data, not the scenarios built by the Central bank out of Blackrock’s estimates (needless to say the Irish banks are calling these numbers absurdly too high!). This is summarized below from the Central Bank’s report.

This does not include Anglo Irish, for which we might add an extra 40bn€ (central bank scenario was at 29bn€), so 80bn€ in total. Focusing on real estate, we get 67bn€ (assuming Anglo is all real estate losses, which is conservative of course).

Going back to our “macro” rule of thumb we adjust for 2010 GDP (Ireland 160bn, Spain 1053bn) and we get the following multiple: Spanish losses should be between 2.2 and 3.3 times losses in Ireland.

This leads ultimately to total losses between 146bn€ and 220bn€, interestingly close to the historical scenario and the stressed scenario from Moody’s. This is a very conservative approach as one must remember that the actual number used by the Irish central bank was only 69% of the Blackrock estimate, due to various (reasonable and realistic) effects.

The three analysis we propose lead to an estimation of recapitalization needs for the Cajas in a stress scenario that is between 60bn€ and 100bn€. The higher number would probably not materialize before two or three years.

3. What does it mean for Spain and the markets?

Assuming the highest number is “achieved” (and we are talking about highly stressed numbers), what would be the impact on financial markets – apart from obvious volatility.

Firstly, we already argued that most listed Spanish banks would be able to face such high levels of provisioning and this is a view shared by capital markets, analysts and rating agencies. Focusing on hybrid debt, any coercive action on bondholders would be useless as it would not save any taxpayer money and it would only prevent Spanish banks from raising equity on the capital markets to cover for the losses. Cajas would need to receive public money, of course, but this would have limited impact as they are not listed and their hybrid debt is almost entirely held by the Spanish retail network. Taking taxpayer money to save taxpayer money makes no sense.

The most interesting impact, obviously, is on the sovereign market and potential ramifications. It would surely trigger volatility on the market and, possibly, though we don’t believe it, an EFSF intervention. But would it really affect debt sustainability of Spain? Evolution Securities has done a great job of data crunching to look at Spanish public finances under stressed assumptions regarding the actual impact of current reforms on public deficits and on bank recapitalization (including restructuring of Portugal and Greek debt). They use a 120bn€ assumption, too high in our view and the numbers are the following.

Even in such a stressed scenario, Spain would be in a much better shape than Ireland, Greece or Portugal and exhibits sustainable debt and deficit numbers. This is probably why Spain is now totally decorelated from the periphery in terms of cash and CDS price variations (see our other research paper on that topic). No mystery here: remember that Spain went into the crisis with only 36% of public debt…

Overall, should stress scenarios materialize at the level of the Cajas, we would not expect a real solvency crisis in Spain as the size of the sector remains manageable for the global Spanish economy.




In the same section