With the approval of the Decree Law 99 of 25 June 2017, the Italian Government finalised the plan to resolve the crisis of Banca Popolare di Vicenza and Veneto Banca (V&V). The Ministry of Economy and Finance, acting on the proposal of the Bank of Italy, put the two banks into compulsory administrative liquidation ("liquidazione coatta amministrativa"). After a long incubation period, during which various attempts had been tried to recover V&V as a going concern, the definitive solution came by surprise as an exception to EU rules that would prescribe resolution as the standard remedy to a bank insolvency, in accord with the 2014 Bank Recovery and Resolution Directive (BRRD).
This informational post aims to clarify the impact of the V&V rescue plan on the financial position of the entities involved, and the final cost to the Italian state. I identify the distinctive feature of the atypical solution chosen in Italy in its slow execution over a long time span, opposed to the quick, "surgical" approach of a resolution process.
The facts and decisions leading to the final arrangement are summarised in a briefing by the European Parliament staff:
On Friday 23 June 2017, the European Central Bank (ECB) declared the two Veneto banks “failing or likely to fail”. The ECB press release indicates that the decision results from capital shortfalls as “the two banks repeatedly breached supervisory capital requirements”. […]
On the same day the Single Resolution Board (SRB) assessed that the conditions for resolution as per BRRD were not met. According to the SRB, while the two banks were failing or likely to fail and no private solution could be implemented to prevent their failure, there was no public interest justifying resolution action. The SRB defends that “neither of these banks provide critical functions, and their failure is not expected to have significant adverse impact on financial stability”. As a consequence, the two Veneto banks had to be wound down under normal insolvency proceedings at national level, under the responsibility of Banca d’Italia, in its capacity as National Resolution Authority.
On 25 June 2017 the two banks were wound down with the transfer of the performing business (performing loans, financial assets, deposits and senior debt) to Intesa San Paolo (ISP) subject to the injection of cash and the provision of guarantees by the Italian government (see below), the transfer of the non-performing portfolio to SGA, the vehicle formerly used for the liquidation of Banco di Napoli, and the bail-in of equity and subordinated shareholders, which remain in the entity into liquidation.
Such a narrative evokes the typical ingredients of a bank resolution procedure compliant to the BRRD: the sale of part of the assets and liabilities of the failing bank and their transfer on commercial terms to a purchaser without the consent of the shareholders fully complies with the definition of the sale of business tool as per Article 38 BRRD. The bail-in of capital instruments is also mentioned.
However, the plan devised in Italy is by no means a bank resolution, both formally and substantially: the SRB dismissed the presence of public interest, the key requisite for putting "significant" banks under ECB supervision (as both V&V where) into an SRB managed resolution; the case was handed over to Italian authorities who, on their behalf, adopted an approach based on national insolvency law with some twists that depart from the spirit of BRRD.
The inspiration from the FDIC P&A + receivership approach
The solution adopted resembles the rescue of Banco di Napoli in 1996, the most severe bank crisis in recent Italian history. In order to make it comparable in an international perspective, I am proposing an analogy with a Purchase and Assume agreement (P&A) used in combination with a Receivership process (as explained in the FDIC Resolutions Handbook). Historically, P&A has been the standard way to address bank insolvencies in the US. In an interesting study on European bank restructuring cases by Dübel, this approach is renamed "Good bank - bad bank model". In a nutshell, a bank insolvency process is split in two and each part is routed to a distinct track:
a fast track for marketing the good business of the failing bank to solid competitors through a P&A agreement; this step ensures continuity of critical functions and credit relationships;
a slow track for the liquidation of the remaining "bad" part of the bank balance sheet by means of a receivership process; the resolution authority acts as receiver and is substituted in legal relationships and pending litigations.
One may easily draw a line around the bad assets that are the primary cause of the bank’s failure: non-core non performing loans (NPL), foreclosed property, or other devalued or "toxic" assets. On the other hand, delimiting the bad liabilities that should remain with the liquidating entity is more of a challenge. Insured deposits and covered or asset-backed securities belong to the good bank, no question about it. But what about uninsured and unsecured funding instruments, ranked senior or subordinated? It depends on the value of the good bank’s asset/liability (A/L) combination. If the good bank has a positive franchise value, the assets may support a higher amount of "made-good" liabilities.
A contribution may be paid to the purchaser in order to compensate excess liabilities. In the setting of FDIC interventions, such a subsidy may be warranted on the basis of a "least cost analysis" demonstrating that the envisaged P&A solution entails a lower charge on FDIC funds compared with insured deposit payoff that would occur were the bank wound down. Within the mission of maintaining public confidence in the nation’s financial system, FDIC will favour solutions capable of minimizing the impact of an institution’s failure on the local economy.
The P&A part of the V&V rescue: the sale of the good business to Intesa Sanpaolo
The decree law 99 includes the sale of the good business of the Veneto banks to Intesa Sanpaolo, which entails a transfer of all performing loans, financial assets and liabilities (treasury, trading and other derivatives), ECB refinancing, senior bonds, and deposits, as well as branches and employees. Such segregated scope excludes NPLs (bad loans, unlikely-to-pay loans and past due exposures), subordinated bonds issued, as well as shareholdings and other legal relationships that the Bank does not consider functional to the acquisition. The following table splits the values in the consolidated balance sheet of V&V at the end of 1Q2017 on each side of the good/bad bank divide.
Intesa has assumed €51.3 billion of liabilities vis-à-vis €45.9 billion of purchased assets, with a €5.4 billion difference (all items are computed at book value). As a compensation for such A/L imbalance, Intesa recognises a receivable from the liquidating (bad) banks, in the form of a 5-year loan carrying a 1 percent interest rate.
The necessary State-aid sweetener
Intesa San Paolo paid a consideration of one euro and received, in addition to the good business of the bank, the following State contributions or guarantees:
the State guarantee on the aforementioned financing of the A/L imbalance; the provisional amount of €5.3 billion may be increased up to €6.3 billion after a due diligence;
the right to re-transfer to the bad banks (at book value) the high-risk loans that will become non performing within the end of 2020; such obligation of the bad banks is State guaranteed up to €4 billion;
a €3.5 billion public cash contribution to its capital in order to maintain a CET1 ratio of 12.5% to risk-weighted assets (RWA) acquired;
an additional €1.285 billion public cash contribution to cover integration and rationalisation charges in relation to the acquisition. These charges include those relating to the closure of around 600 branches and the use of the solidarity allowance mechanism in relation to the exit, on a voluntary basis, of around 3,900 people of the Group resulting from the acquisition;
public guarantees equal to €1.5 billion after tax, in order to sterilise legal risks, plus €0.491 billion to increase allowances for legal risks on the bad banks;
the fully eligibility of Intesa Sanpaolo to use the deferred tax assets (DTAs) of the banks acquired, comprising the potential ones stemming from the carry-forward of huge unused tax losses.
Including such benefits, the following balance sheet of the good bank is obtained.
The P&A deal closed by Intesa should be perfectly neutral on the group’s CET1 solvency ratio. The additional capital charge driven by the increase in RWA is matched by the public contribution. The A/L imbalance financing carries a zero risk-weight thanks to the State guarantee. Defaulting high risk loans will be taken back by the bad banks at book value, with a commitment resembling a free credit default swap, also guaranteed by the State.
One-off restructuring expenses are covered by a specific subsidy, as are legal expenses. Even allowing for additional charges in the integration phase before reaching the economic break-even of the re-branded network, the deal is value creating for Intesa, as is demonstrated by the appreciation in the share price of the bank.
To get the whole picture, one should remember that Intesa and some of its reference shareholders (the Cariplo and Sanpaolo foundations) contributed with more than €1.5 billion to the recapitalisation of V&V through the Atlante fund. Between March 2016 and January 2017, Atlante poured a total of €3.5 billion into V&V’s capital. With hindsight, the State’s contribution to Intesa’s CET1 looks like a sort of indemnification, in the light of the fact that Atlante’s involvement had been strongly encouraged by "someone far from being an usher" (witty allusion to the Government and/or the Bank of Italy made by Giuseppe Guzzetti, Cariplo foundation’s president).
Intesa-V&V vs. Santander-Popular
In the Italian and European media, a parallel has become common place between the sale of the Veneto Banks to Intesa and the previous sale of Banco Popular to the Santander group. Both of them had been closed at the symbolic price of 1 euro. Intesa is reported to have been much smarter in the bargaining with the national Government and banking supervisors: V&V have been taken over clean of problem assets and litigation risk with a hefty public subsidy. On the other side, Santander got the whole Popular balance sheet, including a portfolio of seasoned overvalued property (a legacy of the Spanish real estate bubble) and committed to fill the ensuing capital gap with a share issue, and to indemnify holders of shares and subordinated bonds that have been wiped out in the resolution.
But when the strategic rationale of the respective deals is taken into account, the balance reverses in favour of Santander. The Popular group had a strong national franchise in Spain and Portugal with a leading market position in SME lending (very attractive in a buoyant domestic economy), sporting a best-in-class cost/income ratio. Veneto banks' operations were concentrated in the same region, their brand had been destroyed during years of mismanagement and customer expropriation through mis-selling of shares. They carried opaque risk in their credit portfolio. Cost/income ratio had gone beyond 100% at Banca popolare di Vicenza.
The stock market cheered the Popular deal as well. This fact demonstrates that Santander did not load any dead weight on its shoulders either.
Another common place in the press has been praising the Spanish authorities for the quick and smooth handling of the crisis vis-à-vis blaming the Italians for extending and pretending for years that the two banks were viable only to bail-out them on the brink of a disorderly failure. More on this point on a future post. Let’s go back to our main focus.
The bad bank and its would-be hidden treasure
So far, the V&V plan seems straightforward: Intesa unwillingly takes over the good part of the failing banks' balance sheet, requires coverage of excess expenses and risks, and the State foots the bill. At the end of the story, a classical State bail-out would have occurred. But we have not arrived at the end of the story yet. We still have to read through the intricate part that concerns the bad banks. The former Banca popolare di Vicenza and Veneto Banca go on as entities under receivership. It is in their accounts that the expenses charged to the public budget add up and give the final cost to the taxpayer.
Who manages the NPL portfolio? The role of SGA
In the bad banks' balance sheet we find the remainder of the assets and liabilities that are not sold to Intesa. On the assets side, the main item is the NPL portfolio comprising bad loans (€ 8.9bn Gross book value, GBV, €4.2bn Net book value, NBV) unlikely-to-pay and past due exposures (€ 8.6bn GBV, €5.4bn NBV), NPLs from foreign subsidiaries (€0.6bn GBV, €0.3bn NBV), and the high risk performing loans that may be re-transferred by Intesa upon default until December 2020 (up to €4bn). Excluding the last component, the expected recovery value, before interest and servicing fees, amounts to €9.9bn on €18.1bn GBV, with a 55% expected recovery rate on GBV.
These exposures are maintained at book value. No extraordinary impairment is applied, unlike what would happen in a resolution procedure, where the portfolio would be marked to the market price quoted by distressed debt investors.
The decree law foresees that the NPL portfolio will be sold to SGA, a State-owned non-bank financial intermediary licensed for servicing activity. SGA has a long and peculiar history, having performed a similar role in the rescue of Banco di Napoli. Banco di Napoli sold its NPL portfolio (together with other non-core assets) to SGA in 1996 at about 70% of the respective gross book value. SGA adopted a patient approach. The original 5-year mandate has been repeatedly extended to present days. SGA recruited a qualified team who managed the work-out rather successfully, achieving a 92% recovery rate on the net book value, before operating expenses, that corresponds to around 65% on gross value. In small ticket loans, SGA opted for out-of-court agreements reached through one-on-one collaborative dialogue with borrowers, an approach rewarded by the best performance across all borrower classes. Many big ticket exposures were collateralised by high quality real estate, which facilitated recovery in a scenario of booming property prices. Foreclosed assets were often re-sold with substantial gains.
The role of SGA in the V&V plan will be different in several respects. SGA was the bad bank operating alongside the good bank Banco di Napoli. In the current set-up, SGA should act as a master servicer on behalf of the bad banks under liquidation. As in the previous experience, the good bank staff will collaborate with SGA in collection and servicing activities for a fee. From the legal point of view, the NPL portfolio and other assets will be sold to SGA at book value against a non recourse claim on the proceeds of recovery management, net of servicing fees. Under this arrangement, the recovery risk will stay with the bad banks. The NPL portfolio will be segregated in SGA’s balance sheet, and will not originate any prudential capital requirement on the servicer thanks to a specific provision in decree law 99. In substance, the NPL sale is not a "true sale", but a sort of single-tranche pass-through securitisation where the originating banks remain the sole holder of the securitised exposures.
There are also differences in the expected performance from the recovery process. The V&V portfolio has a dominant share of corporate and SME exposures, for a large part unsecured or secured on non residential property. The average quality of collateral is lower than in the Banco di Napoli portfolio, so the expected 55% recovery rate may prove an optimistic target to achieve. To make things worse, €1.6 billion of gross NPL (almost one tenth of the total) are related to "azioni baciate" (literally kissed shares), i.e. loans directed at the purchase of V&V shares in order to artificially inflate the banks' capital. A great part of these loans may be declared invalid, or are challenged by legal actions for mis-selling the shares.
On the positive side, a public entity such as SGA may exert more pressure on borrowers unwilling to repay but endowed with personal wealth shielded from creditors in tax havens or elsewhere. Italian tax authorities collected information about illegally exported capital in the repatriation initiatives carried on in the previous decade (so called "Scudo fiscale" or "Voluntary disclosure"). Since taxpayers' money is at stake in the bad banks, information on foreign holdings of V&V’s borrowers may be shared with SGA. Such an opportunity should not remain merely anecdotal, because the SGA experience demonstrates that patient and clever negotiation with recalcitrant borrowers may have a payback, and a "stick and carrot" approach even more so. Bank trade union representatives hinted at this happening while presenting a plan (now being dismissed) for securitising failing banks' NPL in cooperation with SGA (see my post, only in Italian).
Bad banks intermediate public contributions
Cash injections to Intesa are intermediated by the entities in liquidation: the State extends a cash advance to the bad banks and the latter transfer the monies to Intesa as a subsidy charged on their income statement. The State gets a senior claim on the assets in the liquidation mass, which is subordinated only to the amount due to Intesa against the A/L imbalance. The amounts paid by the Ministry of finances against the give back of defaulted loans and legal expenses originate a recourse credit that is treated in the same way. Correspondingly, the net costs to the Italian State will be much lower than the nominal amounts of the measures provided.
The bad banks' liability structure and the treatment of shareholders and subordinated debt holders
The liability structure of the banks under receivership is represented in the following table, where the items are listed in reverse order of seniority:
The proceeds from asset collection are allocated according to a simple payment waterfall, articulated in five ranks:
Super-preferred debt related to Receivership expenses financing;
State guaranteed A/L imbalance financing by Intesa and State recourse credit on retransferred high risk loans;
State advances covering contributions to Intesa and other;
Outstanding subordinated bonds;
Values in the table are from the base scenario assumed in the Technical briefing to the Parliament prepared by the Bank of Italy. Apparently, no burden sharing has occurred, since both subordinated debt and equity are still there. This seems at odds with the statement by the DG Competition of the European commission about the approval of the State-aid component of the plan:
The Commission found these measures to be in line with EU State aid rules, in particular the 2013 Banking Communication. Existing shareholders and subordinated debt holders have fully contributed to the costs, reducing the cost of the intervention for the Italian State.
There is a technical difference in the way burden sharing is accomplished: shares and subordinated bonds are not written down, but they remain in the bad bank at the bottom of the payment waterfall. They may participate to the proceeds of asset liquidation, but with a very low probability of getting a payoff and low recovery rate on the book value should that happen. According to our simulations, presented later, sub holders may be repaid in part in the base scenario, if an optimistic recovery target is achieved.
In addition to the unlikely participation to the liquidation payoff, the decree law foresees a restitution mechanism to the benefit of small savers who hold subordinated bonds issued by the two banks, considered as likely victims of mis-selling. Intesa Sanpaolo will allocate €60 million in addition to the funds supplied by the banking system through the Interbank Deposit Guarantee Fund, that is in charge of such mechanism. This provision reproduces the one originally devised to indemnify the subordinated debt holder of the "four banks" (small regional banks in the Centre of Italy) put into resolution in November 2015.
A simulation of the bad bank balance sheet and the final cost to the public budget
To sum up the information about the rescue plan of V&V, I have performed a preliminary simulation of the receivership results, starting from the bad banks' balance sheet. I have added an asset item for "Active legal claims" against the former members of the Board of Directors and Supervisory Board, based on arbitrary personal conjectures.
Consolidated bad banks' balance sheet in the base scenario (€billion)
In the following table the same balance sheet structure is extended with data about the contingent liabilities carried by the State that do not translate into cash outflows and net debt increase at the beginning of the plan.
The "State guaranteed" column reports the maximum amount of the payment that the State is committed to make. The "fair value input" column shows the expected value of the contingent guaranteed payment computed according to the hypotheses made in the aforementioned technical briefing by the Bank of Italy.
Summing the "Total State advances covering subsidies to Intesa and other" (net value) and "Total preferred debt" (State guarantees) we obtain the maximum expense that the State may be asked to cover in a worst case scenario:
Total State advances covering subsidies to Intesa and other (net value) €4.785 billion
Total preferred debt (State guarantees) €12.342 billion
Maximum cost to the Italian State €17.027 billion
The impact on the budget deficit is much lower because contingent liabilities are accounted for at fair value:
Total State advances covering subsidies to Intesa and other (net value) €4.785 billion
Total preferred debt (fair value of State guarantees) €0.7 billion
Maximum cost to the Italian State €5.485 billion
The effective cost will depend primarily on the realized recovery rate on the NPL portfolio. Some preliminary but realistic examples are provided in the following scenario analysis.
Consolidated bad banks' balance sheet with maximum and fair value of contingent State liabilities (€billion)
Scenario analysis of bad banks' cash flows and liability coverage
Just to give some preliminary but realistic examples, I have computed the bad banks' cash flows and their distribution to the various ranks in the liability structure in five scenarios. To keep things simple, I did not simulate the receivership expenses (interest on Intesa financing, servicing fees, governance, operating and legal expenses), apart from an initial endowment.
Scenario hypotheses are summarised in the following table. The first two scenarios keep the assumptions made by the Bank of Italy (the second one adds the extra-financing of the A/L imbalance that may be added after the due diligence). The last three scenarios, grouped as "stress test" cases, assume that recovery rates worsen with respect to the base case, as well as growing amounts of defaulted loans given back to the bad bans, higher activation of guarantees on legal expenses, and lower positive payoff from legal actions.
Source: "base" and "due diligence worst" scenarios are based on the briefings by the Bank of Italy attached to the legislative text. "Stress test" hypotheses are based on subjective plausible conjectures.
The last table summarises expected cumulative cash flows on bad banks' lifetime. Inflows are grouped by asset class. Outflows are grouped by liability rank.
In the first two scenarios the weighted average recovery rate on NPLs is 55%. The receivership closes with the complete repayment of preferred debt principal amount. The residual cash flow should be used to cover interest expenses, servicing fees and other administrative expenses. Under these optimistic assumptions, the cost on the State is zero because the bad banks repay entirely the Intesa financing (without activating the State guarantee), as well as the State funding of contributions to Intesa. After deducting bad banks' recurring expenses, a surplus may remain that could be used to repay a fraction of subordinated bonds. Payments to common equity apparently are not possible.
Moving to stress test scenarios, cumulative inflows decrease and contingent liabilities may be activated. The receivership is no longer capable of repaying preferred debt in full. In the average and adverse scenarios, the State guarantee on Intesa financing is activated and the State steps in as recourse creditor, suffering a loss. Nothing is recovered of the advances related to cash contributions.
The final cost to the public budget is given by the unpaid part of the amounts due to the State, totalling €2.8 billion in the moderate stress scenario (against an average recovery rate of 41%), €7 billion in the average scenario (avg recovery rate 31%), and €12.6 billion in the adverse scenario (avg recovery rate 19%).
Preferred debt coverage and final cost to the Italian State by scenario (€billion)
Source: Our computations on data from Bank of Italy briefings (see previous tables).
I hope this technical notes will help foreign observers understand the philosophy and the execution details of the bank rescue plan adopted by Italian authorities in the case of V&V banks.
My analysis touched upon a number of issues that are hot in current debate. Does the Italian decision represent a serious breach of a key principle of EU legislation on bank crises, i.e. protecting the taxpayer from the cost of bail-outs? Would a straight resolution with senior debt bail-in have been a more effective option? I reserve these and other crucial policy questions for a future post.