Well, the recent Variant Perception seems to have caused the next best thing to a media "perfect storm". Yesterday Iñigo Vega of Iberian Equities went to the trouble of writing a direct rebuttal to Jonathan's report (covered here by Tracy Alloway at FT Alphaville).
Since the situation in Spain is now so serious, and since Iñigo (as he indicates) has been one of the few Spanish analysts who has begun to understand the severity of the Spanish problem (even on his "optimistic" view Spanish banks will have an average of 10% Non Performing Loans in 2010, and what that will mean in the worst case scenario sectors of the Spanish banking industry is anyone's guess) I think his reply is worthy of a rather wider circulation than the traditional analyst distribution, so I am - in the interest of having a clear and open debate - putting the complete text online. Here it is.
The holes In Variant’s research
by Iñigo Vega, Iberian Equities
We have been not particularly shy at the time of pointing up some the structural challenges of the Spanish our two latest long industry reports “Prepare for a hard Landing” we still believe things must get worse before getting better and that there will be clear losers (an not different than in previous cycles. Having Sheet”) contains a number of inaccuracies and misinterpretations Several investors and observers have been coming to us over the last week with specific data requests on issues commented by the report. Find below a summary of our thoughts:
Banks hiding loses- Spain is Japan 2.0- Not a bad start
Variant argues Spanish banks are hiding their loses and this would replicate the long-lasting banking crisis in Japan. To keep things simple, our chart below shows the cumulative change in NPLs (write-offs included) since the outburst of the financial crisis for both the US (at FDIC insured institutions) and the Spanish system. So to say that Spanish banks are hiding their credit loses is a brave statement to say the least. And if they are really hiding loses, in relation to which particular financial system? We are really not aware of any system with a shape in cumulative new NPL similar to the Spanish, at least in developed markets.
€470bn in loans could go bad — Far too simplistic
Variant picks-up a classic wild-card to spice-up the report. Specifically, they say most of the €470bn in outstanding loans to developers/construction (50% of Spain’s GDP) could go bad. The report forgets to mention- however- that a chunk of the €32 bn in outstanding loans to developers does not necessarily involve residential lending but commercial lending (which is relatively safe in Spain , in our view). It does not say either that a not-low percentage of construction activity in Spain involves public works, so a proportion of the construction-related debt (€141bn) should be attached to that public sector accordingly. Also it is worth considering that residential work-in-progress in Spain — one of the biggest contributors to the €320bn figure — is generally collateralized (with Spanish major developers reporting LTV of 50-65% approx). Factor-in these and the final loss on this portfolio should be a fraction of what Variant claims. In our models, we assume a 15% peak NPL ratio on Developers (7.6% in 1Q09) and a 10% NPL ratio on Construction loans (6.7% in 1 09). Another factor to bear in mind is the recent trend in lending to developers (outstanding loans up €5bn in 1Q09 despite the large volume in asset swaps). First the €3bn increase compares with a net fall in the system’s total corporate loans of €2bn. Secondly, we think this shows the high-degree of progress in some of the developments, so assuming a value of zero in many to be- finished units is far too much. Anecdotally, we have also met a large number of decent-size Spanish developers over the last few months that are in good financial shape.
Getting a boost from accounting charges in 1H09- Wrong
Yes, the Bank of Spain changed last July the interpretation of the provisioning rule on some mortgage loans. Now the rule is more in line with the rules applied by most European/US banks (where provisions tend to match the expected loss as opposed to the frequency of losses). However, the measure has had zero impact on the system’s P&L hitherto. The only listed institution that has applied the rule in 2Q09 (Banco Santander) re-classified the release (€270m) as an additional specific provision. Variant claims however- ” the change in rules has allowed Spanish institutions not to lose money this year”.
Dynamic provisioning - Inaccurate
As an starter, the Bank of Spain introduced the generic provisioning system in 1982 (as a 1.5% mandatory provision on new loans) and not in 2000 as Variant claims. Subsequently, the rule was modified in ’87 (at 1% of all new loans). Later in 1991, the provision was lowered to 0.5% for all new mortgage loans (and maintained at 1% for the rest of loans). The current system was fine-tuned in 2004 based on the statistical scheme approved in ’99 (and introduced in July’00). In other words, by ’00, Spanish banks had already stocked €7bn in generic provisions or an additional buffer (as a pct of equity) of 10%. Finally Variant suggests that the benefits of the generic buffer is generally overstated by banking analysts. Well, let’s just say that if the US retail system had stocked a similar amount in generic provisions in the good years, most of its retail institutions would still be at a profit. In fact, Spanish Banks added €12.3bn in excess provisions over the last five years (end ’03-end’08) or 12% of its reported pre-tax profits (0.7% of loans). Under a similar scheme, the US banks would have stocked around US$109bn in excess provisions just ahead of ’07. This would be the equivalent to 1.4% of loans and would more than cover the US$37bn pre-tax loss reported by the system over the last three quarters (see below for data).
Not marking loans to market- Come-on!!!
Variant claims Spanish banks are not marking their loan books to market. Non performing loans in Spain (4.6% of the system’s loans by the end of Jun’09) are marked-down according to different provisioning calendars set by the Central Bank. For non-mortgage loans, NPLs are provisioned at the end of year 2. The majority of mortgage loans (40% of loans or two thirds of mortgage loans) have been - until the BoS made changed the interpretation of the rule - also 100% provisioned by year 2. Only a small fraction of low -risk mortgages (20% of loans) are provisioned according to a long calendar (100% provision by year 6). By international standards, Spain’s provisioning calendars are quite strict especially considering >60% of loans have a mortgage collateral.
If what Variant really suggests is marking to market the whole loan portfolios the same way CDOs are, then Santander’s recent tender offer for its own bonds is self explanatory, in our view. Basically, Santander is offering to buy some of its own bonds at a 15% discount to par value. And some of these loans are of the lowest credit quality within Santander’s Spanish credit portfolio. If we exclude the quite particular UCI vintages, the discount offered by Santander goes down to 10%. This discount suggests, 1) Santander is simply taking advantage of price inefficiencies in illiquid securities, 2) Santander is happy to buy most of its own outstanding paper at a 10% discount, a fraction of the discount seen in the markets a few months ago. Hence it would have been non-sense to mark-to-market the whole loan portfolio using illiquid bond prices- some of the lowest credit quality - as a market reference. Or if marked-to-market then, they would need a huge upgrade now.
Not marking asset swaps to market given conflict of interests – Additional comments are required.
Variant also notes that because of the strong interaction of property appraisal companies with banks, these are not marking-tomarket the assets bought from developers properly. Variant is mixing things here. First of all, we estimate that Spanish financial institution tend to swap assets at an 20- 30% discount to the initial appraisal value. Subsequently it applies a (mandatory) 10% provision to that value (20% in some cases). That provides a mark-down of up to 37% to initial appraisal values. Yes we agree that there could be some downside in some properties given the poor environment but the difference should not be huge from current levels.
On the conflict of interest between banks and appraisal companies (something we stated 3 years ago already), Variant forgets to add that most of the property transactions in Spain are not supported by an official appraisal. In fact, believe it or not, only 20-25% of the residential homes in Spain (vs. 50% in the US, for instance) have been mortgaged. So values provided by appraisal companies are important price indications, but still there are a large number of transactions where appraisals companies have nothing to say and where free demand/offer reigns.
Offering 100% 40 years LTV loans. A bubble like practice – Not a great issue, in our view.
We think this refers to Santander mainly. As hypothetical shareholders in Santander, we really would not have a huge problem with this practice at this stage as it actually adds granularity to the existing asset portfolio (“better to own 1,00 different interest-yieldingmortgages than a building plot”). Also note Santander is selling at a large mark down relative to initial appraisal values. These markdowns provide a lot of protection to the lender in terms of the potential severity. Bear also in mind that delinquency rations in the core residential mortgage market (that is residents only) is still low, at around . In other words, we would not expect a large expected loss in this portfolio.
Edward Hugh is only able to update this blog about once a month at the moment, but he does have a lively and enjoyable Facebook community where he publishes frequent breaking news economics links and short updates. If you would like to receive these updates on a regular basis and join the debate please invite Edward as a friend by clicking the Facebook link at the top of the right sidebar.