14/06/2017 1 Banco Popular 1. What were Popular’s problems? The initial problems The initial problems of Popular were rather well known, including from the supervisors. They can be broken down into two issues: A much too large real estate book (37bn€), a legacy from the financial crisis and the result of foreclosures on non-performing loans. This book was decently performing: Popular had set up a dedicated department to wind it down and to optimize sales, had a partnership with an experienced servicing firm and was selling assets at a 2% premium to book value. On top of the real estate book, the bank had a large NPL ratio (14.7% in the latest EBA data – the most reliable source, we believe.) However, in April 2016 the Bank of Spain introduced a new circular on provisioning, with, in particular, new rules on appraisals, collateral haircuts, etc. With the limited disclosures available (EBA, company reports) we had estimated the shortfall at 6.7bn€ and the company booked 5.7bn€ of provisions at FY2016. In Q1 2017 the bank booked an additional €506m. A mortgage book that was exposed to the mortgage floor litigation risk following the ECJ ruling on full retroactivity. This led to a 229m€ provision in Q4 2016 but also put pressure on the NIM, by far the biggest share of Popular’s earnings. These problems were mitigated by the following considerations: The bank had a very successful and very profitable core SME commercial franchise; The bank had been able to raise equity successfully to compensate for some of the provisions (including recently in May 2016.) The crisis The crisis started to unfold when the bank indicated that its internal audit had uncovered (a) a provision shortfall and (b) financing granted to clients during the latest capital increase 1 for 205m€. This led to media frenzy and speculations that large additional provisions were required on the real estate book and that the management should envision a new strategy. The management put on a dismal performance in combining various potential strategies without giving a clear direction to the market: 1 This in itself is neither illegal nor a wrongdoing but it should lead to CET1 deductions.
12
Embed
Banco Popular - AXIOMaxiom-ai.com/.../06/Banco-Popular_Axioms-Analysis.pdf · Banco Popular 1. What were Popular ... a legacy from the financial crisis and the ... on the real estate
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
14/06/2017
1
Banco Popular
1. What were Popular’s problems?
The initial problems
The initial problems of Popular were rather well known, including from the supervisors. They
can be broken down into two issues:
A much too large real estate book (37bn€), a legacy from the financial crisis and the
result of foreclosures on non-performing loans. This book was decently performing:
Popular had set up a dedicated department to wind it down and to optimize sales, had a
partnership with an experienced servicing firm and was selling assets at a 2% premium to
book value. On top of the real estate book, the bank had a large NPL ratio (14.7% in the
latest EBA data – the most reliable source, we believe.) However, in April 2016 the Bank
of Spain introduced a new circular on provisioning, with, in particular, new rules on
appraisals, collateral haircuts, etc. With the limited disclosures available (EBA, company
reports) we had estimated the shortfall at 6.7bn€ and the company booked 5.7bn€ of
provisions at FY2016. In Q1 2017 the bank booked an additional €506m.
A mortgage book that was exposed to the mortgage floor litigation risk following the
ECJ ruling on full retroactivity. This led to a 229m€ provision in Q4 2016 but also put
pressure on the NIM, by far the biggest share of Popular’s earnings.
These problems were mitigated by the following considerations:
The bank had a very successful and very profitable core SME commercial franchise;
The bank had been able to raise equity successfully to compensate for some of the
provisions (including recently in May 2016.)
The crisis
The crisis started to unfold when the bank indicated that its internal audit had uncovered (a) a
provision shortfall and (b) financing granted to clients during the latest capital increase1 for
205m€. This led to media frenzy and speculations that large additional provisions were required
on the real estate book and that the management should envision a new strategy.
The management put on a dismal performance in combining various potential strategies without
giving a clear direction to the market:
1 This in itself is neither illegal nor a wrongdoing but it should lead to CET1 deductions.
14/06/2017
2
The management was very late in announcing that the bank was officially for sale and
instead it sounded interest from local buyers. Foreign banks and investment funds were
effectively shut out from the process.
The management initially rejected a capital increase and then “soft hired” advisors to raise
equity – but this was too little too late. Institutional buyers were struggling to understand
why they should invest equity in a bank when no competitor seemed interested, despite
the huge potential synergies, which they obviously did not have.
The management had no communication strategy and was merely reacting to the news
flow. They were not even able to organize an investor day to present a comprehensive
strategy.
Creditors, such as the large AT1 holders, were totally kept in the dark and were reassured
that it was business as usual. The largest AT1 holder was not able to access management
to discuss strategic options.
As we explain below, the authorities justified the resolution by a liquidity crisis. Hence, the
ultimate downfall of Popular demonstrated another weakness of the bank: despite a high
Liquidity Coverage Ratio (“LCR”) the bank’s deposit base was not granular enough, with several
very large tickets. This is partly explained by the SME franchise of the bank but also, we think, by
some of its political connections. The numbers are not public but it is very likely that, compared
to peers, the bank had a smaller share of guaranteed deposit (i.e. below 100K€.)
2. Why was the decision so shocking to the market?
We believe the market was shocked by different aspects of the crisis which we discuss here.
The swiftness of the SRB’s actions
We believe a lot of investors were shocked by the celerity of the SRB’s decision, especially
outside of a weekend. However, we understood from various conversations we have had with
banks, including Santander, investors and journalists, that the FROB actually took control of
Popular earlier, maybe as soon as Friday night, and organised the sale to Santander on Tuesday
night. This is not unusual for the resolution of a bank that usually takes place over a weekend.
True, the management was still contemplating various capital actions and even the possibility of a
sale, but it seems that liquidity concerns left them without the time they required.
All junior bonds got wiped out with no creditor hierarchy
Some investors have expressed disbelief at the 0 valuation for Tier 2 bonds. It is true that the
provisional valuation, that was made by an independent advisor, in only a few hours we
understand (!), had a very wide range of values, and that the higher bound (-2bn€) was
suspiciously close to the amount that allows the resolution authorities to effectively wipe out
junior creditors. Still we are not that surprised: as long as the bank is declared in resolution,
bondholders have basically two protections:
The valuation of the bank should show that losses to be absorbed are higher than the
amount of the junior bonds available to absorb them.
The no creditor worse off principle implies that creditors should not receive less than
they would have received in a normal bankruptcy.
14/06/2017
3
However, we have always expressed serious doubts on the strength of the protection offered
by these two principles. A Tier 2 or AT1 tranche in a bank’s balance sheet will generally
represent between 1% and 2% of RWA, if not less. Considering an average 40% RWA
density, this means the size of such liabilities is between 0.4% and 0.8% of a generally very
complex balance sheet. In spreadsheet modelling, when you have to value a bank in a
weekend, it is very clear that a number like that is well within the margin of error. The
valuation range for Popular was 6bn€, i.e. 3.8% of total assets. What this means in practice, is
that a zero recovery is a logical and credible scenario both for AT1 and Tier 2, and possibly
non-preferred seniors, unless there is a political will to protect some tranche of the balance
sheet or there is a buyer willing to make whole these investors. In other terms, when it comes
to bank risk, the probability of default is much more important than the loss given default.
Some investors have also been shocked that conversion terms included in the AT1
documentation were not respected (i.e. there was no conversion at the price given in the
prospectus) and that Tier 2 were converted despite no such clause in their prospectus. We
believe this criticism is unmerited: the conversion powers of the resolution authorities are
very clear in the BRRD and the bond prospectuses are largely irrelevant here.
The market believed the PONV was still distant
The exact trigger of a bank’s non-viability remains a grey area in European legislation and the
case of Popular came as big surprise to most investors. True, the CET1 ratio was low, but it was
above the MDA threshold for distribution restriction, the supervisor had allowed for a recent
AT1 coupon payment, and the bank was clearly working (although not very efficiently) on a
strategic plan that included, to the very least, asset sales likely to generate 200bps of CET1.
Potential additional LLPs were being discussed, and rumoured to be below 2bn€, which seemed
manageable for a bank with a pre-provision profit around 1bn€/year and a 10bn€ book value.
Crucially, and unlike many problem banks in Italy or elsewhere, the core franchise of the bank
was very profitable and delivering a good ROE, suggesting that there would be an appetite for
capital raising post dilution.
Moreover, the possibility to convert AT1 bonds, especially the one with a 7% CET1 trigger,
seemed to allow for an additional 1.25bn€ CET1 buffer (i.e. 200bps CET1) should additional
provisions lower the CET1 ratio below 5.125%. Clearly, the most bearish market view was that
the bank was viable, possibly after 2bn€ of additional LLPs and an AT1 conversion, and it seems
that some distressed investors would have been interested by the bank at a low price-to-book
valuation (maybe 0.2x). We can think of a few banks in Europe that have a lower “viability”
(starting with the four Greek banks or the two Venetians) and that have not been put into
resolution by the SRB.
Last but not least, even after taking into account the “abnormal” (see below) provisions taken by
Santander (7.2bn€) the equity of the bank remained positive.
However, the main reason for the resolution, as explicitly stated by the ECB, was the liquidity of
the bank, not its solvency: “The significant deterioration of the liquidity situation of the
bank in recent days led to a determination that the entity would have, in the near future, been unable to pay
its debts or other liabilities as they fell due.” Obviously, this remains a valid ground for determining that
14/06/2017
4
a bank has to be resolved but it came at a surprise to the market as we discuss below into more
details.
Liquidity was not perceived as a substantial risk for Popular
Since 2011, liquidity has become a non-issue for banks. Indeed, authorities have taken successive
decisions that have allowed banks to access almost unlimited amounts of liquidity provided they
are solvent:
The MRO facility of the ECB is now unlimited in amount provided that collateral is
available;
The new LTRO/TLTROs facilities have given banks access to very large amounts of
medium term funding;
New collateral rules have allowed banks to dramatically expand the amount of available
collateral;
Several legislations have allowed banks to issue covered bonds collateralised by a wider
range of assets;
The ECB and national central banks have showed their willingness to provide
emergency liquidity (ELA) even for the most distressed banks (90bn€ for Greek banks!)
with even larger eligibility for collateral;
Governments have granted guarantees on senior bonds, including very recently for
Italian banks, and with the blessing of the European Commission;
Deposit guarantees have been harmonized in the EU and there is now certainty that
deposits under 100k€ are fully protected.
This is probably why most investors felt that liquidity was not a risk for Popular and were
shocked that it turned out to be the sole motivation for the resolution– especially with a very
decent 146% LCR ratio and with a large portfolio of liquid government bonds.
However, it appears that a quite remarkable series of events happened, as reported by the press in
the past few days.
The FROB and ministry of finance disclosed that the bank had lost approximately
18bn€ of deposits in a couple of days.
From the discussions we have had with two of the three largest banks in Spain, we
understand they saw combined inflows of only 1.5bn€…leaving a 15bn€ gap even with
the assumption that the other commercial banks also received similar inflows.
The Spanish press reported that the deposit outflows were largely due to withdrawals
from public clients: local authorities (Canaries allegedly withdrew 600m€), social
security funds (allegedly withdrew several billions), SAREB (allegedly withdrew 600m€),
etc. Allegedly the ICO also had a large deposit balance with Popular and we understand
that it was withdrawn too.
According to the press, the Bank of Spain requested a 90% haircut on the bank’s
collateral. Allegedly, Popular offered 40bn€ of collateral and was offered…only 3.5bn€
in funding in exchange, which is quite extraordinary considering the ECB’s general
documentation where the maximum haircut is 65%.
14/06/2017
5
There have been conflicting reports on the bank’s access to the ELA: according to
official statements the bank had limited access to the ELA, but the press reported that
according to internal sources at Popular, the bank was denied any ELA.
All in all, it seems the liquidity dry up was rather unusual and it is quite possible that the
government did not feel the need to prevent it – or to stop public institutions from withdrawing
their money. Clearly, this demonstrates a couple of important points:
The LCR calculation is somehow flawed because it assumes standard withdrawal
rates on deposits (these rates are different for corporate, retail, etc.) but any coordinated
decision to withdraw very large deposits by a few actors could be enough to destabilize
rapidly the balance sheet of a bank;
A much diversified deposit base is much stronger than a concentrated one. In particular,
we think the share of guaranteed deposit will become an important metric when
assessing the liquidity risk of bank. Obviously not all depositors will act rationally and
some could be influenced and scared by media frenzy, but in a world where deposits
below 100k are fully guaranteed and where the BRRD grants a senior ranking to
guaranteed deposits, we think these deposits should be stickier, even in distressed
situations.
The auction was surprising
Another reason for the market shock was that many global, large, investors were still waiting to
hear from the investment banks mandated by Popular to reach an opinion on a potential
recapitalisation or even an outright sale. Clearly, the feeling was that the new strategy was still a
work in progress. However, we understand from conversations with the banks themselves and
press reports, that only two banks were allowed to submit a bid for Popular and that they had
only 11 hours to do so. Many investors have expressed surprise at such a closed-door auction and
do not understand why they were not consulted. This could also be a consequence of the
management’s actions as it chose to consult only a handful of Spanish banks for a potential
merger. These banks were already in the process of a due diligence and had more information
than the rest of the market.
The procedural requirements set out in the BRRD are rather loosely drafted although the general
principle is that, if a sale is organised, there should be a minimum of competition. Clearly, here, if
the reports are correct, it was the absolute minimum (two banks.)
Again, this demonstrates a few important points for future cases:
During a resolution process, it is very unlikely that institutions without a deep
knowledge of the bank will be able to make a price (for the bank or some of its
portfolios)
In practice, this means that the sale of business tool (or an outright sale) is unlikely to be
the preferred resolution tool, as in most cases there will be no investor with a clear view
on the bank. However, should special circumstances apply, such as the one with
Popular, i.e. an ongoing sale process, then the potential buyers will be in such a strong
14/06/2017
6
position that they are unlikely to make a generous bid and to offer any protection to
shareholders or junior bondholders.
3. Does this crisis show a significant shift in the SRB/EU’s resolution
policy?
Clearly, the first and foremost concern for investors should be the read-across of that resolution
to other banks. This raises two very different questions: (i) has the SRB changed its stance
regarding resolution and (ii) are there any other banks significantly at risk? We deal with the first
question below.
All considered, we think there has been no shift in the EU or the SRB’s stance towards
resolution:
- In terms of valuation, all banks that have been put in resolution have had negative
valuation estimates to justify full or partial write-downs on subordinated debt.
- In terms of burden sharing, since the EC’s decision to change state aid rules, a full
contribution of junior debt (write-down or equity conversion) has been considered a
prerequisite to any state aid and we think it is fair to assume that in most resolutions a
similar approach will apply.
- Popular’s case has been different from other situations where equity conversion had
some value (e.g. Bankia or MPS more recently.) We believe there are two reasons why
this did not apply here
o a buyer for the whole bank was found, taking on board all senior liabilities,
and this was obviously the safest option from a systemic risk point of view.
Based on the (obviously disputable) valuations made by the advisors, this
could even be considered a good deal for Popular’s creditors as any negative
valuation below -2.3bn€ could have justified a haircut on senior debt.
o The main issue (see above) of Popular was a liquidity issue. Converting the
subordinated debt into equity would have not contributed in any way to
restoring liquidity.
- The market focused probably too much on the bail-in tool, but resolution authorities
can use a variety of tools – and have used them in the past! The SRB used the "sale of
business tool" (Article 38), the write-down tool (Article 59) but not the "bail-in" tool
(Article 43) in particular on senior debt (probably because Popular had very little
senior bonds.)
In other terms, we think the SRB has not changed its approach but was confronted with a
situation where the only way to stop a senior debt default was to sell the bank to a larger bank –
at whatever price that bank was willing to accept.
14/06/2017
7
The true and significant shift was in Spain’s approach to the crisis: there was no political will
whatsoever to grant liquidity measures to Popular or to provide capital. In that context,
where there is no time, the resolution authorities, which are independent and have no power to
grant liquidity measures, have basically only two options: a default on senior debt (haircut,
moratorium, etc.) or the sale of the bank.
4. What is the read-across for other banks?
The first and immediate read-across concern is regarding asset quality. Indeed, immediately
following its acquisition for 1€, Santander announced 7.9bn€ of additional provisions, bringing
real-estate loan coverage to 75% and real estate assets coverage to 65%. Non-real estate loans
coverage was increased to 56%. Benchmarking these very high coverage ratios to many of the
smaller Spanish banks – or Italian, Portuguese, Greek and Irish banks for that matter – would
raise considerable capital concerns.
The truth, as always, is more complicated. Indeed, Santander itself acknowledges that these
coverage ratios are very high. They were chosen to allow a full sale of the entire real estate
portfolio within the next 18-36 months to distressed investors looking for a 20% return!
Under IFRS rules, yields used to compute a loan’s coverage ratio should be the loan’s yields, i.e.
rarely above 5%. Even regulatory overlays do not lead to such pharaonic returns.
Moreover, coverage ratios are not arbitrary numbers, they depend, among other things, on the
type of NPE (forborne, past due, NPL, etc.) on the type of debtor (corporate, financial or
household) and on the collateral value.
So the bottom line is that Santander’s choice of coverage ratios should definitely not be
used as benchmarks for other banks.
Still, we believe there is a substantial non-performing asset overhang in the EU - even if this is
certainly not new. We show below the results of the detailed benchmarking analysis of NPE that
we have performed on a wide sample of EU banks with the latest EBA Transparency Data
available (this is the only dataset that includes consistent and harmonized data in the EU.)
14/06/2017
8
Source: Axiom AI, EBA
Apart from the Greek and Cypriot banks, we can see obvious shortfalls for some midcap Spanish
and Italian banks that we think will need to be addressed at some point. Interestingly, in the few
cases where supervisory authorities have forced banks to take extra provisions or where the
banks have decided to do it themselves, our estimates have been very close (e.g. Monte dei Paschi
with 4.9bn€ or Unicredit with 13.2bn€.) Any bank which would see its CET1 ratio go below 9%
(obviously an arbitrary threshold) after taking these extra provisions should only be considered
with caution.
This study does not include other NPAs, i.e. foreclosed real estate, but the situation is very
different for those. Spain has been “penalized” by the quality of its enforcement procedures: in
many EU jurisdictions, banks are struggling to repossess the collateral on their loans and,
logically, own very few real assets. On the contrary, in Spain, NPE have been going down
because banks have been able to write them off against the collateral (which is good) but now
have large real estate books. It is not easy to harmonize the disclosure of all banks on this,
especially smaller ones, but we believe EBA data provides, again, the best disclosure. In this data
set, real estate is booked as NCO, i.e. non-credit obligation. Apart from a few odd cases (e.g. VW
Bank which obviously has many cars…) the 7 banks which have the largest share of RWA in
NCO are all Spanish, as shown below.
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
CY
16
CY
17
CY
18
CY
19
ES4
3
ES4
4
ES4
5
ES4
6
ES4
7
ES4
8
ES4
9
ES5
0
ES5
1
ES5
2
ES5
3
ES5
4
ES5
5
ES5
6
GR
71
GR
72
GR
73
GR
74
IE76
IE77
IE78
IE79
IT80
IT81
IT82
IT83
IT84
IT85
IT86
IT87
IT88
IT89
IT90
IT91
IT92
IT93
IT94
PT
113
PT
114
PT
115
PT
116
PT
117
PT
118
NPE provisioning shortfall as % RWA
14/06/2017
9
Bank Share of NCO RWA
Banco Mare Nostrum 24.8%
ABANCA Holding Financiero 20.9%
Banco de Sabadell SA 19.5%
Unicaja Banco SA 19.1%
Banco de Crédito Social Cooperativo SA 18.3%
Liberbank 17.2%
Banco Popular Español SA 16.8%
Source: EBA, Axiom AI
This is why we focus here on those Spanish midcap banks: Abanca, Cajamar, Ibercaja,
Kutxabank, Liberbank and Unicaja (BMN is likely to be absorbed by Bankia so should not be
studied as a standalone entity anymore.) Clearly, as always in real estate, the type of asset is a
major driving factor for the coverage ratios. Land, properties under construction, finished
properties and residential properties will not have the same risks and the same coverages. Using
the split of each bank’s book (data as of H2 2016) we use two scenarios to benchmark the banks:
one in which each bank is benchmarked to the maximum of its peer group and one in which each
bank is benchmarked to the average of its peer group plus one standard deviation. The provision
shortfalls, expressed as % of RWAs, are given below.