Where Does Italy’s Bank Recapitalization Stand?

There are broadly speaking four approaches to a banking system saddled with lots of potentially bad loans:

1) Forbearance. Allow the banks to hold the questionable assets on their books at a mark that is higher than they would get if they sold the assets in the market, and potentially higher than the long-run recovery.

2) Private recapitalization. First force the banks to mark the assets down (and potentially sell the assets) and take a hit to their capital. And then force the banks to rebuild their capital by raising it privately, or by converting some portion of their debt into equity. Junior bonds (subordinated debt) for example. And if there are not enough junior bonds, senior bonds…

3) Public recapitalization. Force the banks to mark down their bad assets (and potentially sell them to the market or to a bad bank) and rebuild the banks’ capital with government funds. This can be combined with a private recapitalization—subordinated debt can be converted into equity, with the government providing additional equity.

4) And finally, liquidation of the bank—which requires disposing of the banks’ good assets as well as their bad assets.

In practice, financial troubles are usually solved with some combination of forbearance (markets do overreact, and forcing banks to mark their books to fire-sale prices would lead to large and even excessive recapitalization needs), private recapitalization, and public recapitalization. Banks in the U.S., for example, were not forced to mark their “hold to maturity” book to market in the U.S. stress test back in 2009, but they were forced to raise private capital—and to retain/take public capital if they could not raise private capital.

Up until now, Italy has relied primarily on forbearance—allowing the banks to hold a large stock of non-performing loans on their balance sheet at a relatively high mark.

Partly because of its large legacy stock of retail holdings of both junior and senior bank bonds.  

Absent new private investors willing to step in and buy (and recapitalize) weak banks in the way Santander recapitalized Banco Popular in Spain, a private recapitalization means large losses for retail investors, not just institutional investors. Retail investors who bought the bonds well before Europe agreed to the Bank Recovery and Resolution Directive (BRRD), back when Italy typically protected holders of bank bonds from losses.

And also in part because Italy didn’t want to add to its stock of public debt.

Italy is now willing to use the public sector’s balance sheet for recapitalization, which has allowed a bit of forward movement.   But public injections of capital need to be consistent with European banking union and state aid rules. Europe seems to have created binding rules that limit national sovereignty (though probably only to some degree) before it created well-funded European institutions for sharing the fiscal risk associated with public injections of capital and deposit insurance.*

I have long worried that the solution to these intersecting constraints will still be a bit too much forbearance, with too many banks relying on favorable marks on the risky exposures to avoid some combination of private recapitalization (bail-in of junior and senior bonds) and public recapitalization (bailout). And I have also believed Italy should get a bit of flexibility in exchange for putting more capital into its banks, as the sustainability of Italy’s public debt depends more on having a banking sector that can finance growth than limiting the size of the check that Italy’s treasury will need to write to rebuild banks’ capital.**

So I wanted to review where I think we now stand, with a focus on balance sheet repair rather than on the debate over how to interpret the BRRD’s rules for precautionary recapitalization. 

-------

With the help of my colleague Emma Smith, I’ve tried to sketch out the aggregate capitalization of the Italian banking system at the end of the year—using publicly disclosed information from the Banca D’Italia and the listed banks. There is no guarantee this is completely right. We aren’t pros and certainly lack the detailed information available to the regulators.

Italian banks, indeed European banks, generally operate with a pretty small amount of equity relative to their total assets (we used total assets here, not risk-weighted assets). At the end of 2016, Italy’s banks look to have had about €187 billion of equity capital, supporting €3.8 trillion in total assets. Those assets include something like €340 billion in Italian government bonds, €1.7 trillion in good loans, and roughly €349 billion of non-performing loans (NPLs).

The non-performing loans are on average marked at about fifty cents on the euro: the worst of the NPLs—the sofferenze—have a lower mark, other troubled loans have a higher mark. So, net non-performing loans were about equal to the system’s capital going into 2017—that isn’t a great place to be, but it also isn’t a disastrous place to be.

(A technical note we’ve had to use a slightly old estimate for core tier 1, from the end of June 2016, as a starting point; the estimate is calculated from information in the Bank of Italy’s financial stability report.***) 

But there subsequently has been a bit of progress.

Unicredit raised €13 billion in new capital, which we’ve added to the mid 2016 equity estimate to give €200 billion total equity. (The €1 billion of capital raised in the merger of two Milanese banks—Populare di Milan and Banco Popolare—is captured in the June 2016 data.)

Unicredit also added €8 billion in provisions against its NPLs. For the system as a whole, provisions are up by €11 billion since mid last year. And total non-performing loans are down by €7 billion. This means that, all up, banks’ NPL exposure is down to €173 billion, from €191 billion in June 2016.

Monte looks set to mark actual bad loans down to around 20 cents on the euro (from a current mark of 35 cents) so it can unload these into the market (in a securitization vehicle, with the senior tranche guaranteed by the government) and fill the resulting capital hole through the conversion of junior bonds to equity and the injection of about €6.6 billion in public capital through the precautionary recapitalization mechanism. (According to Bank of Italy estimates, €4.6 billion in direct state aid plus €2 billion to buy out the equity of bailed-in retail investors.)

Monte accounts for about €46 billion of Italy’s €349 billion in NPLs. As the below table shows, €30 billion of these are bad loans—the worst category of NPLs. €15 billion are other NPLs. Moving €30 billion off its books will meaningfully impact the total: Gross NPLs will fall to €319 billion, and net NPLs from €173 billion to around €162 billion.****

And negotiations are ongoing over two Venetian banks with a combined €19 billion in bad loans (offset in part by €8.5 billion of provisions). These banks appear to have about €2.6 billion in core tier 1 capital at the end of 2016 and roughly €1 billion in outstanding subordinated debt. (CET1 estimates are from the banks' end-2016 financial reports—here and here, with almost all the capital stemming from the injection of funds from Atlante. It isn't clear if the end-December numbers include the last €938m Atlante capital injection/ Jonathan Algar has a higher number for tangible equity.)

It now looks like a proposed deal to inject an additional €1 billion in "private" capital in order to make the two banks eligible for a precautionary recapitalization from the Italian state (as a precautionary recapitalization can only use state aid to cover future losses not existing losses) has fallen through, and Intesa will instead buy the banks’ good assets for a nominal sum—and presumably also take over most of the two banks' liabilities.   

But as always the split between the good bank and the bad bank is critical. Watch where the senior bonds go, and where bonds guaranteed by the Italian government go. Writing off the two Venetian banks' equity and the sub-debt would leave net NPLs of something like €7 billion (a very rough estimate), and someone will have to bear the risk that the ultimate recovery is less than that.  

The likely loss to Italy's taxpayers here isn’t all that large in the grand scheme of things, so we are inclined to be a bit flexible here and to interpret the European rules in a way that protects the senior bonds and limits systemic risks.   But not everyone agrees, this isn't a done deal yet.

Together with the Monte restructuring, creating a bad bank for the bad Veneto bank assets would bring gross NPLs in the system down to around €300 billion and net NPLs  down to just over €150 billion.

Finally, the really small cooperative banks will be organized into a couple of groups and the groups will be subject to a stress test (a good move). We wouldn’t be surprised if this leads to a rise in troubled exposures in these groups.

Is that going to be enough? Maybe. Italy’s economy is now recovering, and that helps. NPL formation is slowing.

But we’re not sure—we would like to see a bit more capital in the system. What stands out, at least to us, is the set of institutions with more net troubled exposure than capital (i.e. a Texas ratio of over 100 percent) and still relatively high marks on the valuation of those exposures.

Hence our preference for adding public capital to more institutions, even if that means more losses for their subordinated debt holders. 

Monte actually had a decent amount of provisions against its most troubled exposures. Its mark was high but not super-high. It’s certainly lower than most of the other banks in our sample. Monte’s problem was the sheer size of its troubled loan book (almost as many troubled loans as good loans)—the cost of a troubled loan book rises significantly in the event of “funding” stress.   That is, we suspect, part of what gave rise to its precautionary recapitalization need.

The two Veneto banks have, proportionately speaking, more performing loans than Monte. Their problem is the combination of high marks on their NPLs and very little capital. Carige (Genoa) and the merged Milanese banks also don’t have as many troubled loans relative to their total lending as Monte, but again are carrying those exposures at relatively high marks.

Italy’s bad loan exposure exposures are actually quite diverse. They are mostly on loans to businesses—and have different collateral backing. That differs a bit from Spain, when the worst exposures were all to large real estate developers, and thus backed by various forms of real estate collateral. There could well be good reasons for significant differences in the marks on various banks’ NPL portfolios.

But there is also uncertainty on the ultimate return—uncertainty that would be mitigated by holding enough capital to allow the bank to absorb any downside risk.

We consequently would be more comfortable if total equity capital in Italy’s banking system was significantly north of €200 billion, and net non-performing exposures were substantially below €150 billion. Based on what we can see developing now, we do not think Italy will quite get there.  

Italy's economy is now finally growing: it should take advantage of the eurozone's current recovery to build up some real capital in its banking system in case the good times do not last.

One final note: Bank of Italy data shows €152 billion of retail bond holdings (junior and senior) of the end of 2016. It also shows that €102 billion rolls off by end-2019, so in two and half years total retail holdings of bank bonds could be down to €50 billion. The retail problem could largely solve itself—making the application of new bail-in rules easier with time.

DISCLAIMER:  We have tried our best, but we aren’t bank analysts. No doubt we have a few things off, especially as we are chasing a moving target.

* Here it is worth noting that Spain’s 2012 recapitalization limited the losses of holders of the preferred equity of the banks that received public equity injections. The preferreds took losses, but were not completely wiped out. The losses on the holders of the sub-debt were also limited. And senior bonds did not take any losses. That recapitalization is broadly viewed as a success today, though it arguably was a bit too narrow. (Banco Popular probably should have been forced to mark its real estate loans down then and to have taken public capital. It got a fairly generous mark in Spain’s 2012 stress tests.) The combination of the BRRD and ever-tighter interpretation of state aid rules has significantly limited the use of public capital in bank recapitalizations. Italy now is being held to a more demanding standard than Spain was in 2012, with far more burden-sharing by retail and other investors in the banks junior debt.

** In Italy’s case the bank-sovereign doom loop isn’t symmetric. The banks’ €340 billion in exposure to Italy’s sovereign is far larger than any reasonable estimate of the cost to the government from bank recapitalization. And the solvency of Italy’s government ultimately depends on Italy’s ability to grow.

***The report gives a Texas ratio of 101 percent as of end-June 2016. Using data on non-performing loans and provisions allows us to back-out the amount of equity. The updated Texas ratio given in the table is calculated using more recent (end-2016) data on NPLs and provisions. Again, no guarantees that we’ve got this exactly right.

**** The change in net NPLs is equivalent to the value Monte is carrying the bad loans (€10.8 billion). They’ll also take a €4.7 billion hit on their capital, reflecting the difference in the price at which they’re carrying the bad loans (35 cents) and likely purchase price (20 cents).