Month: November 2014

My response to David Einhorn’s comments (Part II)

Sorry for the delay.

Same structure as in Part I. Warning: this is a long post.

David Einhorn

  1. Relating to the fall in funding costs:

The 2.04% is the cost of new time deposits in August 2014. This is not the average cost of deposits in the Q2 2014 financial reports, which was 2.62% according to the same presentation (and 2.71% in Greece time deposits). This cost will reduce down to the new cost of deposits over the 1-2 quarter duration of the time deposits. So even without any additional lowering of deposit pricing you would still expect to see a near 60bps improvement. This is obviously very material given there are €29.6bn of time deposits (the rates are also coming down on savings and sight deposits but this is much less relevant). The 60bps is worth close to €200m of additional annualised PPI, compared with the €1.2bn annualised reported PPI in Q2 2014. How low can deposit rates go? Time will tell of course but rates in the other periphery are around 1% and the management teams of the Greek banks are clearly trying to reduce them towards this level. I would suspect another 50bps.


Good point. Time deposits are between 50bp-75bp in Austria, so 1% in the periphery means deposits are priced almost like in the economically stronger parts of Europe – you decide whether that makes sense. If no shocks happen between now and 2017, PPI could indeed get a nice boost from that.

But: how likely is this to happen, if your France thesis proves correct? Do you think that an imploding France will have no repercussions on the periphery (or, indeed the whole Euro zone) countries?

David Einhorn

  1. Relating to the rising cost of long-term Greek Government debt

This is a classic confusion between correlation and causality. Both deposit rates and GGB yields were falling up until recently. Since then there has been a sudden rise in GGB yields. Why? I believe the primary reason is that markets have started to discount political risk in Greece (this has also impacted equities as well, of course) ahead of a likely full election in 2015. These GGB’s total €30bn nominal in debt and have a market value of around €18bn. They are primarily held by hedge funds and other similar investors. There is no obvious reason why the trading of these investors in Greek Government debt should influence the behaviour of the average Greek depositor (particularly since depositors that left the Greek market to deposit elsewhere in Euroland have not returned to Greece). And so far this is what we have seen. Deposits increased at the banks in October and both banks recently confirmed to me that they saw no change in consumer behaviour (as of last week). Deposit rates in November are still going lower.


Sorry, I did not make myself clear enough: there is no such thing as causality (if A then B) in economic phenomena. This is not (Newtonian) physics! What we have instead are complex “relationships” that can run both ways: there could be a bank run on deposits caused by rumors that negatively affects the sovereign as its bailout capacity is doubted. Alternatively, you could have doubts about the sovereign affecting the soundness of banks causing a run. Since there are continuously changing leads, lags and even coincidences between the two variables, it is not possible to make quantitative statements about the relationship whatsoever.

This, however, doesn’t mean they have nothing to do with each other as you seem to think, just because the bulk of the tradable debt is held by non-banks.

First, the exposure is still there and it will roughly double for Piraeus as DTAs are converted into government debt to better comply with Basel III in the future (see below) – I am sure the Greeks will manage to pull-through this stunt without increasing the official deficit. Second, Piraeus is heavily dependent on the interbank market for funding (up to 25% of its BS) and this is definitely more directly linked to GGB yields: lending banks can now lend to the Greek state at higher yields (and no equity requirement whatsoever) than in the summer and the ECB has to ask for more Repo-collateral due to the fall in price. If things stay this way, the treasurer of Piraeus will sooner or later have to start balancing between the increased cost of interbank funding and the deposit rates he offers.

I would state, then, that it is highly unlikely for both variables (yield and deposit rate) to show a diverging trend over a considerable period of time.

And the political risk that is currently discounted won’t go away for a long time in Greece – it is real.

David Einhorn

  1. Relating to lack of loan growth

I do not rely on loan growth in my analysis, but on cost reductions and funding costs. However it is reasonable to suggest that GDP growth and loan growth do tend to correlate quite well and so it is quite possible that some loan growth might return to the market, as the economy recovers.


You might not rely on loan growth for your medium-term earnings forecast, but you implicitly rely on growth assuming a multiple on 1.5 in 2017. According to the modified DDM (bot the worst model for valuing banks I think) shown in the last post, a no growth bank with a cost of equity of 12% (reasonable for Greece) should trade at anywhere between 1.17 – 1.25 x book depending on whether you assume a ROE of 14% or 15%. Without growth, you either need a lower cost of equity or a higher ROE.

David Einhorn

  1. Relating to Table 2.2 (Developed markets debt breakdown)

This table seems to show something quite different to what you think. Whilst the 304% is correct the loan growth, or lack thereof, is likely to come most from the private sector (households and non-financial companies). Here Greece has debt to GDP of 129% (with the Government adding 175% to get to the 304%). This is actually the second lowest figure in the table, with only Italy lower at 125%. The US is higher at 160%. Leverage at non-financial companies is only lower in Germany and household leverage similar to most of the core European nations (France, Germany, Austria, Belgium). So, there is at least some room for private sector loan growth.



First, the distinction between private and government is somewhat arbitrary, as high government debt needs to be serviced by higher taxes which impedes private sector growth. It also leads to crowding out as the government commands more and more resources, i.e. there are links between the two. Note that crowding out in the modern monetary system is not via interest rates, as classic economic theory would suggest – the link between demand for debt and supply of loanable funds (saving) not being direct anymore due to credit-money creation – but via the increase of prices of factors of production (rents, wages of qualified personnel…) which prices the private sector out.

Second, as I argued in the post Greece should not be compared to developed markets, but to EM most of which have significantly lower debt loads. Greece’s institutions are closer to those of, say, Hungary or Serbia than to Germany and the US. And the institutional backdrop matters more in banking, than, say, in retailing. In a country in which a court needs years to enforce a mortgage on a house, it is questionable whether you should offer this product at all. Similarly, countries where most earn a significant part of their income on the black market most customers are not bankable (hence the need for decentralized Microfinance) as incomes are not verifiable. The reason why the banking sector in Greece is so much bigger than in other EMs, is because its banks could hold lover liquidity buffers (and hence create more deposits) as they had direct access to hard currency – this will revert to the mean.

If this doesn’t convince you then ask yourself why there is no loan growth at Piraeus bank. The equity seems to be there (more on that below). Marginal funding at around 2% is low, as we have seen. Loaning out that money at 5-6 percent earns a nice spread of 3-4 percent. These are not usurious levels, but still the loan book is shrinking. Something doesn’t add up here, if you ask me.

David Einhorn

  1. Relating Piraeus’ topline growth potential:

Your analysis also excludes the other major source of gains in PPI, which is cost reduction. They have €152m of additional synergies as part of their restructuring programme. This is to come primarily from branch closures and employee cost reductions through redundancies. Their medium term targets assume they will go further than this and I also think that plausible given how over branched Greece still is given the # of banks.


There is not much I can add here. You have more experience than me judging management synergy estimates. This chart, from the Blackstone study you mentioned arouse my interest.


It shows Piraeus’s acquisition activity over the past two and a half years. As a result of six acquisitions/mergers 58 percent of its current loan exposure was not underwritten in-house and I suppose each institution had its own underwriting standards. Worse, it also probably means that each institution had its own core banking systems that need to be harmonized. This can be very costly and entails a high degree of operational risk. Are these costs already included in their synergy estimates? If not, you can safely deduct a high double-digit million figure from their estimates.

David Einhorn

  1.  We do not contend that the bankruptcy law change has anything to do with business foreclosures. Further, as we stated in the presentation, this will be a much bigger benefit for Alpha Bank than for Piraeus.

Ok. From the Blackrock piece I take away that the foreclosure moratorium is related to primary mortgages, as in Hungary – just as I expected. I want to point out that the end of the foreclosure moratorium has been postponed to the end of the following year every time in the past three years. It would not surprise me if that happens again. This is related to the institutional environment I wrote about above.

David Einhorn

  1. Relating to collateral:

Mortages are only a small part of Piraeus. The collateral is not PP&E. You ask a bunch of questions:

“What type of do they count here? I hope it is not property, plant and equipment. How do they arrive at the figures? Transactions? Hardly possible in a depressionary environment. Estimated values (Appraisals)? If yes, how old are they? Are the values reported nominal values or haircut values? And how much is collateral worth in an environment where enforcement is legally difficult, anyway?

Why don’t you try to answer them?

Blackrock published a 173 page piece of research on all these issues (they also came up with stress test cumulative loss provisions that are very similar to those that the ECB stress test calculated) and it is available online. It describes answers to all of these questions.

Let us take mortgages to start with. Piraeus, as per Blackrock in March 2014, had €17.3bn in mortgage loans. This represents 340k loans with an average balance of €51k, average coupon of 3.4%, average LTV of 71% and a weighted averaging seasoning term of 54 months (either since modification or since origination). 79% of these loans were current, 6% delinquent (90-359 days past due) and 15%. Values are updated on an annual basis through indexation to PropIndex. Blackrock then hired real estate agents to perform “drive-bys” to check to provide independent valuations. These valuations were 13-14% lower than on the books (this had strong correlation with loan size, and here Piraeus is lower than average so it’s possible that their adjustment factor might be lower). Let’s assume you haircut the collateral by 14%, then total mortgage coverage is 99% (cash and collateral). And this would be before ANY impacts from strategic defaulters or any recent modifications of loans ahead of the change in legislation.


EUR 17bn is about 25 percent of the loan exposure – not that small I would say. This exposure is affected by said law and here I ask the general question: What’s a LTV worth if you cannot foreclose?

Since real estate is by far the biggest collateral in the SME (bulk of NPLs) and the CRE segments, conclusions from this section extend to them as well.

But let’s not philosophize and have a look at the data (quality)!

Blackrock, as I understand, used banks’ LTV values and it updated them relying on a property price index. In the case of Piraeus bank most RRE loans have been on the books for about 6 years (54 months seasoning). Now, the first and most important question is this: were the LTVs originally based on purchase transactions or, did they have to be guessed at the origination date as is the case if you take out a loan on a property you already own? If it is the latter, it is very likely that the values were inflated from the start, just as occurred in the US during the subprime crisis (HELOC loans). Unfortunately, Blackrock doesn’t provide us with this important information. The fact that for each borrower there are 1.4 x as many mortgages (page 21) suggest to me that a fair degree of speculative activity was going on. Also the relatively small loan size (50k) hints at home equity withdrawals.

The “drive-by” valuations were done in a non-representative sample size and Blackrock duly warns us to extrapolate the results. So we are stuck with very coarse data (apart from that, “drive-by” valuations are garbage anyway).

Since Blackrock doesn’t really answer any of my questions, we can use some deductive reasoning to get a better feel for the situation.

I start with the apartment price index published by the Greek central bank (PDF here). Not surprisingly, the price of the average apartment in Greece has fallen by some 35% over the past five years. If correct, it doesn’t seem that bad considering the depression in Greece – I have seen worse in other CEE countries. With this additional information, it is now possible to back out the implied LTV at origination. Since most loans are not older than six years, we can assume that not much of the original loan balance has been paid back – mortgages start amortizing later. Starting with Blackrock’s LTV of 71, assuming that 10% of the loan balance has been paid back so far on average and dividing the denominator by 0.65 I get a LTV of 0.51 at origination. Wow!

This means that the average Greek mortgage debtor came up with 50% equity at the origination of the loan – and this at a time when mortgage loan growth in Greece was running at between 20-30 percent per year!

Of course!

From this recent Bank of Greece publication I could get a better feel on the Greek real estate market. There I learn that the home-ownership rate in Greece is 80% – much higher than in the US and similar to the situation in other periphery countries. This means that the market is illiquid and index values are based primarily on surveys of banks and real estate agents, i.e. we can safely assume the RE index data themselves are biased upwards. The chart below tells the whole story:


Turnover has been constantly falling since 2005 hitting 48k transactions in the whole of Greece in 2013. Given that Piraeus has 340k mortgage loans outstanding, of which about 20% are either defaulted or delinquent, means that about 68k apartment units have to be foreclosed (if it were legally possible, that is, and twice as much if you include the modified ones). It would take Piraeus bank alone two years to sell its mortgage collateral – not counting other banks that are even more into mortgages and not counting new (re)defaulters.

What does this tell us?

It tells us that the Greek banks have a “collective action” problem: if one starts liquidating RE prices will collapse and losses increase. It also means that a 14% haircut on this collateral is a joke.

The ECB AQR website is a treasure trove. There you can compare banks on a country by country basis. And it confirms what I suspected: at 14% (that’s end of 2013 figures, now it is about 17%) coverage of RRE NPLs it is among the least conservatively covered banks in the EU. For instance, almost all Italian and Spanish banks whose RRE portfolios were selected had significantly larger coverage than Piraeus. Is this plausible, given that the situation in Greece is by far the most severe?

If even the RRE market is illiquid, you can bet that the CRE market looks even worse.

David Einhorn

  1. Then you ask “If collateral covers the net loans, why is the NPL ratio rising and standing at a record 38% of all loans, why are they not getting rid of these loans?”

I think your question misses the point. Given that most potential purchasers of books of NPL’s do not possess servicing facilities in Greece, and that there was no law for reclaiming collateral in mortgages and an insufficient law for restructuring corporate loans, who would buy these loans and why would you want to sell them at the appropriate risk adjusted return these buyers would demand? In fact Piraeus has just got to setting up its restructuring division (2k employees were trained over the summer) and will take most of the next year from now to attempt to restructure its NPL’s. They can perhaps be criticised for taking too long to have got this far but then there has been bankruptcy, sovereign default, various cost restructuring plans, recapitalisation on the market, Blackrock I and II, the ECB AQR and stress test, running a business etc. Plus you are seeing a new law that will deal with corporate restructurings passing through Parliament now, which will enable rapid restructuring of Corporate NPL’s with agreement from 50% + 1 shares of the debt defaulted. Given that the banks will always hold this much they now have far more control over the process (this should reduce the process from 2 years to under 1 year).

Further, the first signs even in consumer credit are positive. Piraeus offered restructuring to 20k customers who had defaulted and had been threatened with legal action. They offered forgiveness of between 30 and 50% of the loan if they repaid the rest. 10k started paying again. There is evidence everywhere of people who can pay, at least a bit, and are not.

Also note that in areas where collateral is consistent with international peers some transactions have taken place, ie shipping, where Alpha have completed a few transactions in the space.


Well you would like to sell them to improve your liquidity situation and free valuable equity which also costs you money, wouldn’t you?

In order to appropriately provision under IFRS you anyway have to discount expected proceeds, just like a potential buyer would. If properly provisioned (big if) the discrepancy between book and market value should not be that high. Technically speaking there is a difference since under IFRS you discount at the original contract rate and a buyer would discount at the prevailing market rate. I admit that in Greece’s case the difference could be material. So what? Shouldn’t we be interested in getting a good estimate for BV? Under non-market assumptions even Lehmann’s Level-III assets would have been correctly valued at book…

And it is not entirely true that there are no potential buyers for these loans. About 9bn (RRE, SBP, Consumer) of the defaulted 28bn are very small ticket (<50k) loans. Nobody is going to fly in and build servicing facilities for them. After calling and threatening the clients, these loans at the end of the day are sold to guys in black leather jackets with large, barking dogs who then pursue whatever they can get. I know that such “services” exist in countries surrounding Greece. And given the healthy hooligan scene I would be surprised, if Greece were an exception.

The fact that they haven’t had a proper workout department doesn’t surprise me the least. In boom times this is a drag on performance unnecessarily distorting the cost to income ratio. Happened everywhere. But starting to build one six years into a depression is shocking. There is no excuse for that!

Who has talked to the clients over the past six years? Who has made sure that they have paid the annual insurance contract on the pledged collateral? Is the collateral still there, at all? Covenant monitoring? Who has made the modification decisions of which there are many according to Blackrock?

I think I know the answer: the sales guy was in charge, i.e. the one who advocated the loan in the first place. He has been the client’s first point of reference during the depression – not exactly best practice. He is also the one whose job is most likely to be cut due to synergies…

I do not know whether the law as you describe it will change that much. For most of the business NPLs Piraeus will be the only bank in the capital structure anyway, since SMEs are the biggest default segment. More importantly, the biggest impediment to quick restructurings, apart from a shortage of workout professionals, will be the lack of speed of the Greek courts (institutions) – regardless of the law.

Given that they had provisioned 73% on consumer loans they could afford to offer this deal. Note, however, that even if the clients have started paying, this is not your normal performing portfolio as the bank doesn’t have any behavioral score data on the client, i.e. it is impossible to assess the riskiness of the written-off (?) consumer book. It also increases the likelihood of strategic defaults in the future, since clients know that it “pays” not to pay.

The shipping book is due to its international nature not really useful in assessing the riskiness of Greek loans. If I were a Greek bank manager trying to cook the books, this is last segment where I would try. First, values can easily be benchmarked due to independent global pricing. Second, given the small relative size of the shipping NPL (1.2 bn) it doesn’t really pay-off.

David Einhorn

  1.  Relating to your conclusion that NPLs from say 2010 have lost value:

Most corporate customers are “cash and carry” businesses. So what happens when you get consumer spending growth and economic growth? Surely you get a much healthier corporate sector? As can clearly be seen from the Blackrock report companies fell in trouble due to large declines in EBITDA, in many cases leaving margins –ve or net debt to EBITDA of more than 8x. This is obviously why provision levels here are high at 67% in cash. But there could well be a significant change here if the cycle turns and growth returns.


Yes, if you get (substantial) growth, things could change – that’s why I say yours is a growth story. And it’s not just real economic growth you need, but especially nominal EBITDA growth across the economy which is more closely tied to money (loan) growth. And it’s not all “cash and carry”: according to Blackrock a huge percentage of businesses has receivables to the tune of 50% of sales!

And yes, falling EBITDA is how companies usually get into trouble. But: even if growth materializes, it doesn’t affect all sectors equally. Only cyclical companies will come back, but a lot of these loans are malinvestments (that’s the cost of a credit bubble) and supported business models that should never have existed, or at least should not have existed at that size. Think of all the construction companies that must have nicely benefited from the Athens Olympics and the great Greek road construction works of the past two decades, the will not see a return of their business that soon (sovereign debt again).

According to their latest presentation cash coverage of business loans stands around 50% – not 63%. Apart from that I found the following chart from the Blackrock piece interesting.


This chart summarizes Blackrock’s findings for all large Greek banks in the SME segment, the largest default category at Piraeus. About 31% of borrowers could not pay interest, whereas 11% can service interest payments, but could nevertheless be classified as defaulted as the likelihood of a principal repayment is very low (EBITDA 1-1.5x). Adding up these two categories yields 42% which is about the NPL ratio at Piraeus and other Greek banks on their business loans. And now, please have look at the rightmost column: on a whopping 32% of loans there is incomplete information, presumably missing financials such that the respective EBITDA ratio could not be even computed. Needless to say, it is not usually the good clients that stop submitting data to the bank.

Another interesting tidbit: about 42% of loans classified as performing (and not provisioned for) have a median debt/EBITDA multiple of 6.2x – surely limiting things like capex and the like. Economizing on maintenance capex destroys business value over time – that’s why workout speed is crucial!

David Einhorn

  1.  Relating to Figure 8. The AQR adjustment chart:

Modifications were made to Piraeus totaling €2.2bn. This is a downwards adjustment in book value that they apply for the purposes of their analysis in the AQR and stress test. This analysis will not be reflected in the financial reports of the banks, and that probably gives a good sense of whether they are merely prudent or actively punitive (if they were prudent you would expect the banks to immediately adopt them). This largely relates to use of collateral vs cash in dealing with corporate exposures. In the event that a corporate is a “going concern” (defined as net debt to EBITDA of <6x) then only cash flows can be considered, and if the corporate is a “gone concern” then only collateral, with significant haircuts applied (mainly to real estate). This is a sensible measure to take for a stress test, but this is not the reality in which the banks are dealing with day-to-day in an attempt to get paid on their loans.

I think there is a misunderstanding. The AQR is used to get a better estimate of asset value which then forms the basis of the stress test. So first you deduct the AQR results from book value, then the stress test. You do not apply the stress test irrespective of the AQR results as this would lead to double counting.

I have read chapter 4 (describing the AQR) of the ECB manual and must say it is sensible. Its purpose seems to have been to get a better feel for the market value of the loan book – exactly what we are also interested in. Now, a multiple of 6x might sound very harsh, especially in the current (Wall Street) environment. But, is it really? Compared to transaction values on heathy corporates with nice margins yes, but for corporates under stress there is the real risk of losing costumers/vendors, negligence of important maintenance etc. The AQR multiples did not fall from the sky, they apparently were taken by observing transaction multiples for distressed sales in the different segments (that’s what they say on page 134).

The AQR if done properly (big if), it is methodologically a good approximation for the true book value, since through its methodology it evades problems that plague IFRS provisioning such as forbearance etc. Still, it was the local regulator who was in charge of this exercise and there is a conflict of interest – surely no one wants its banks to fail. Those that failed were known to be in trouble. Further, given the desolate data quality at most Greek banks it is questionable whether they could have performed the AQR properly. Missing data should have been fully punished according to the methodology in which case the 2.2bn sound too small.

David Einhorn

  1.  Relating to Figure. EBA Stress test results.

This is the Basel III fully loaded ratio under the adverse scenario in the ECB/EBA stress test. Obviously you would not want to own the Greek banks if you believed in the assumptions backing the adverse scenario, which include:

A cumulative GDP decline of 2% from 2014 through 2016

14% further price hits to real estate, and 38% to investment property

17% more NPL’s (as a % of gross loans) from 2014 through 2016, versus 1.5% in H1 2014

A decline in net interest income (driven by NPL’s and higher deposit costs) to 45% lower than the current run rate (which is more severe given that we have already earned higher rates in 9M 2014)

I don’t believe that any of these things are likely.

But, there are also further problems in this analysis.

The balance sheet date they have used for this is year-end 2013, so of course this ratio does not include the capital that was raised in 2014, for a start

Basel III in principle is a very good thing. The aim was to link capital ratios to capital that is actually loss absorbing and not just accounting book equity. So no more goodwill, or embedded value of life operations, or DTA’s etc etc. This has hit the Greeks, and other periphery banks, who amongst other things have lots of DTA’s as they have had lots of losses. So for Piraeus (and Alpha) the bank capital ratio in the chart above would NOT include the Greek Government Preference shares, nor any DTA’s, nor Pillar I bonds (subsequently sold) nor any capital raised

The ECB has, apparently, informed the banks on how to restructure their DTA’s into DTC’s so they will be loss absorbing capital from a Basle III perspective. Reflect that, and the other changes I talked to, and the Basle III ratio will begin to be very similar to the CET1 ratio. This is still low – but then of course I do not expect the adverse scenario


My argument has nothing to do with your or my expectations. Capital is supposed to be there for the UNEXPECTED losses – expected losses are supposed to be covered by risk costs (remark: you can evaluate yourself whether the average mortgage rate of 3.7% at Piraeus covers the risk of these loans). The stress tests are a way to estimate unexpected losses and asses banks capitalization on this basis. Now, we can have a long debate whether it is theoretically possible for bureaucrats to estimate unexpected losses or not, but that’s what they are doing – and it has implications (see next point).

It seems the measures you mention, were indeed not included which renders the results somewhat worthless. Just want to repeat that a conversion of DTA into DTC raises Sovereign risk for Piraeus substantially.

David Einhorn

  1.  You write “Piraeus bank will have to come up with substantial equity in the next five years – it will be a long time before Piraeus can grow its loan book substantially, let alone pay dividends even if profitable. “

This is simply incorrect. That is not the ramification of the stress test.


Not so sure. While EBA has not yet made it explicitly clear what purpose these stress tests ultimately will have, it is safe to assume they will follow the way of the Fed which uses similar tests to govern capital return at banks. Results of these tests do not necessarily have the same consequences as violating regulatory capital, but they still can limit management’s dividend payout policy by prescribing higher capital buffers (Basel III).

Going forward, there are going to be annual stress tests where these things are evaluated on a regulatory basis and if I interpret these results correctly, it will be a long time before Piraeus will be allowed to pay dividends. Of course, I could be overly harsh if the results do not include measures taken in the meantime, but then again nobody knows how the results would look like if everything is included, at least I could not find it.

David Einhorn

  1. Relating to the margin of safety.

I think you are conflating our guess of future value with margin of safety. Margin of safety has to do with downside protection from current levels. Today, Piraeus trades at significant discount to book value, even with full AQR adjustment.

As for the upside, you still assumes that PPI does not go higher than €1.2bn per annum, whereas there are obvious gains through operating expenses and deposit costs (already achieved at today’s market rates) that easily add over €0.35bn. This is with no recovery in volumes or other revenues or any further reduction in deposit costs (every 10bps is worth another €30m) or any more cost saves.

Time will tell how much upside can be achieved. But, in our experience taking a static analysis of banking results at the end of a six year local depression is hardly reflective of what can happen in a recovery.

Even using your view of the Czech as the proper comparison with a 14% ROE, the recovery value for Piraeus is far above current prices.

Overall, I hope you don’t mind my commenting. I enjoy a thoughtful response and a healthy debate.

Time will tell who is right and we will all see how this plays out.

Best de


Piraeus has BV of equity of EUR 9.2bn as of mid-2014. Deducting the EUR 2.2bn from the AQR gives me EUR 7bn as better indicator of “true” BV. At a market cap of EUR 7.2bn, I do not see much downside protection given the risks that I have described.

The fact that the depression has lasted for six years doesn’t necessarily mean it’s over. There is no magic number after which economic growth mean reverts – think Japan. Economic growth only occurs under specific conditions, in human history it is the exception rather than the rule. It is certainly not as easy as Keynesian eggheads seem to think, i.e. by adding historic productivity to population growth. It is sad to see what passes as science these days.

I personally think that the large (and building) NPLs in Greece (and not just there) are an impediment to strong growth, despite the reforms taken. It tells me creative destruction has not happened yet, but this would be the topic for an entire post or study.

I hope my remarks have been helpful and sincerely hope your thesis plays out and you make money on this investment, despite my scepticism.

Best regards,



My Response to David Einhorn’s comments (Part 1)

What a surprise: David Einhorn responds!

As some of you might have noticed, David Einhorn has commented on my post and responded to my arguments. This is a big surprise and I have no clue who told him about my blog. Needless to say, I feel flattered that my analysis aroused his interest. From what I can tell, he seems to have written it himself and, given that it is a lengthy two-part response, he must have put some time into it. Wow!

Although I had planned not write extensively about Piraeus bank again, since I have already spent quite some time on a stock I do not plan to own, David’s response compelled me to have another close look at the matter. To make a long story short: it is always a pleasure to engage in a discussion with a smart and dedicated opponent.

In order to make it easier to follow the arguments, I have copied in his comments and italicized them before responding. Comments which yield similar answers have been grouped together in order to save space. I have split the post into two parts.

David Einhorn:

  1. Greek Employment is improving. It was 28% in Sept 2013 and been flat or better every month since and is now below 26%
  2.  Greek GDP just did turn positive in Q3 (+1.7%). The 6 year recession appears to be over. While growth forecasts across Europe have disappointed, Greece has had a series of beats. Coming off such a depressed level, it is quite possible that there will be more beats in the future. But, like any forecast, time will tell.

Viennacapitalist (VC)

There is obviously no point disputing the numbers and I too have stated that I believe the Greek economy has gotten more competitive at the margin. Since we are grown-ups we are not going to discuss about GDP-forecasts. Just do not expect any miracles: Greece lives in an economically weak neighborhood (EU, Balkans and Middle East) something very few people talk about, but no less a real problem.

David Einhorn

  1. I never said that Greece’s high debt load is “not an issue”.

I would argue that it is much easier to have all your debt basically with a single entity. Particularly when you owe that entity so much that there is little to nothing that they can do but make it easier for you over time – or of course risk that you suddenly default on all of it and leave a giant black hole in the EFSF. You can see how this has already played out with the bailout loans already having reduced rates of interest and longer maturities applied. Plus in any event of default on public debt the ECB will obviously be involved, as like last time, as they own it too.

  1. Greece does not have a material budget deficit.


Ok, you did not say “not an issue,” but you fail to discuss the impact of the sovereign burden on the Greek economy – at least, that’s what I inferred from the absence of any slides on that topic. To be fair: I might well be you discussed it at the conference and did not put it into the slides.

The fact that the Greek government owes most of its debt to a single (non-economic) entity (actually it is a few supranationals that are involved here) certainly seems favorable at first, as they have an incentive to “pretend and extend” and make uneconomic loans just as you describe. And Piraeus bank through its cosy relationships with the Greek central bank certainly can profit from that – just look at the negative goodwill, which, if correctly booked, means taxpayers have handed over quite some assets. But once we factor in the second and third round effect the picture looks different: uneconomic loans enable the uneconomic structures to persist (think: Japan’s Zombie corporates) hampering “creative destruction!” Without creative destruction there cannot be growth, something which the whole developed world is being reminded of, after the bail-out orgies of the past few years. And economic growth is what you need for this investment to make sense in my view.

Loans by supranational organizations (IMF, EU) have historically provided local demagogues with a convenient scapegoat thereby increasing political tensions. Not for nothing will you find very few adherents of IMF programs in countries that have “enjoyed” these “uneconomic” loans. And this is not without substance: usually a few profit, whereas the rest suffers higher prices as these loans are inflationary at the margin. Objectively, the effect of debt programs on growth is questionable as well: there is empirical evidence that IMF programs have, ahem, a mixed record creating sustainable growth. Most fail. Bureaucrats rarely know where to invest. The same is true, by the way, for foreign aid (I very much recommend this book).

Indeed, Greece’s primary balance is forecast to be positive by the EU Commission – my primary reference mentioned in the post. Nothing to quibble about that. But I observe the political tensions are nevertheless growing between the troika and the government. This week’s Bloomberg article  hints at heavy fights behind the scenes about the true deficit figure (emphasis mine),

(…)Troika representatives are furious because the Greek government has failed to come up with any concrete measures to plug the fiscal gap since euro-area finance ministers warned earlier this month about a lack of progress in Greece meeting its commitments, one person said. With the government in Athens refusing to concede there is a funding hole, the standoff means Greece may miss a Dec. 8 deadline for agreement on the steps required to unlock the aid and what comes after it, both said (…)

Seems there is a gap after all. Well, if even they cannot agree on whether there is a deficit or not… 🙂

David Einhorn

  1.  You ask how the debt services is so low absent a restructuring. Here’s how:

Debt service is obviously a reflection of interest payments paid on the debt and he doesn’t seem to have thought about that – I never suggested the total debt amount was decreasing as it is not. Greek debt used to primarily GGB’s paying a market rate of interest back in a time where interest rates were not zero. Now the vast amount of the debt is borrowings owed to the EU and the IMF. Total debt is €320bn. Of that, €154bn is owed to the EFSF and the IMF as part of the second bailout. This carries no cash interest payments at all until the decade long grace period is over in 2022. The €53bn EU loan related to the first bailout pays interest at Euribor + 150bps, down from the original loan fee of Euribor + 400bps. In total Greece owes €245bn to the EU, ECB and IMF and pays cash interest on this of just over €2.5bn. They also pay zero interest on their €15bn of GTB’s (covered by ECB guarantee). The €30bn of restructured GGB’s pay 2% interest. All of this data is readily checkable on Bloomberg and the internet

You suggest 4.3% of GDP interest expenditure, which implies 8 billion of payments on 320 billion of debt, or 2.5%, yet we can see above that only a small portion of the debt carries and interest cost above that rate. I suspect the EU forecast may include accrued interest on the 154 billion bailout loans that are adding up during the grace period. Even this is questionable as the EFSF loan does not have a fixed rate of interest chargeable as far as I can see. In any case, I was referring to cash debt service costs, and I stand by my presentation.


Again, the 4.3% were NOT suggested by me, I took them from the EU Commission’s autumn forecast, as this seemed a natural way to start. I personally very much appreciate that you made your own calculation in a bottom up fashion. Still, given the complex and sometimes (deliberately) opaque nature of the bail-out deals it is easy to miss something.

I say this not because I have found your reasoning faulty, but because your slide (debt-service as % of GDP) puzzles me: How can your forecast for 2014 be substantially lower than your own 2013 figure? Your 2013, figure (4%), by the way, is perfectly in line with the EU Commission’s data. Why the divergence between your data and the EU data from 2014 onwards? Why this dramatic change from 2013 to 2014 in your OWN data? As far as I know there has been no dramatic new bailout benefit effective 2014 – the bulk of the restructurings took place in 2011/12. This is why the interest burden on your slide decreases substantially from 2011 (7%) to 4% in 2013 – fine! However, why there should be another substantial decrease from 2013 to 2014 I do not understand, but maybe there is an explanation  – the 250 bp. decrease on the EUR 53bn that you mention can’t account for that as this amounts to “only” 0.6% of GDP.

On the EU Budget figures: GDP accounting is usually a far cry from double entry bookkeeping and is mostly cash flow based which is exactly why it is so easy to fiddle with the numbers (ask GS). Even if the difference were due to accruals as you suggest, this still would not alter my conclusion that the debt load of the sovereign is still among the highest in the Euro zone: accrued interest increases the debt burden just as much and liquidity constraints for the sovereign are anyway not an issue with the ECB lending freely against Greek government bonds and even without collateral – think ELA.

David Einhorn

  1. We do not believe that the Greek banks we own have high direct sovereign exposures.


Agree! In my original posts I came up with EUR 3 bn. of direct sovereign exposure – not much. However, I was primarily referring to the indirect (growth and political) risks arising from the sovereign debt overhang, as described above and I believe they are important to the story.

Note, however, that the direct sovereign exposure could increase by another EUR 3 bn. if the DTA are converted into a loan to the sovereign, as is being planned by the periphery countries. Curious if this will be deficit increasing, clearly the sovereign debt would rise. This measure is necessary because under BASEL III DTAs will be deducted from capital and the Greek banks will need to come up with more – see discussion on Basel III in the second part of my response.

David Einhorn

  1. Regarding the issue of “moat”: I presented a thesis for 2017. I think it is extremely easy to suggest that there will not be any foreign banks in Greece again by then. And presumably if there are overseas banks that enter Greece, it is because the thesis has already worked very well. It’s hard to see why they would decide to allocate capital to Greece again unless the incumbents were earning outsized returns and they thought there was the opportunity to do the same.


Agree: in 2017 there will be little foreign banking activity in Greece. However, a moat is a long-term concept. Sustainability is important. A few years of outsized returns will not do much to your P/B multiple if they are perceived to be cyclical. In 2017 people better believe that the moat is permanent, otherwise no premium. Last time I checked, most commercial banks in the Euro zone traded at a discount to book value…

I have to concede, however, given that nowadays companies are rewarded by buying-back overvalued stock, it might well be that short-terminism has won out and such considerations do not matter anymore…

David Einhorn

  1.  Neither Dexia nor RBS were bailed out due to failures of their local banking franchises.

Dexia was a liquidity problem, where it lost wholesale financing and included massive Sovereign bond exposure to peripheral Europe into the peak of the crisis. It’s core Belgian operations have earned excess returns throughout.

RBS had its own problems (ABN deal, US real estate, over stretched balance sheet). It’s core UK business earned an ROE between 16-18% from 2004-2007 followed by 11% in 2008, and 3% in 2009, before returning to the teens in 2010. By 2012, with less competition, it hit 24% and was 30% in H1 2014.

We aren’t expecting anything that great in Greece, but it is illustrative of concentrated banking markets can yield excess returns, especially in a post-crisis recovery.


I know it is popular and best practice to distinguish between solvency and liquidity problems. However, I have found this distinction too fuzzy for practical purposes. Would Dexia have earned outsized returns if it had better matched its balance sheet? Probably not! Would the “business model” exist at all? Hard to know for outsiders. For a lot of these municipal finance companies (like Dexia) the “raison d’etre” was levering tiny spreads with maturity mismatches. Looks great as you need little (regulatory) equity and as long as the yield curve is steep. If there is a run, however, they fail that rapidly that not even the ECB can get together in time to save them – doesn’t sound like a good business to me. Certainly, the capital losses suffered by the shareholders during the bailout need to be deducted from the other banks’ ROEs if we want to talk about the “banking system” profitability of a specific country from a top-down perspective as you did in your presentation.

Now, I had to make myself familiar with RBS. For starters: RBS is a collection of different businesses and as a group its performance has been disappointing. Berenberg Research forecasts a negative (-1.1%) Return for 2015. In the table below (Berenberg research) you have all ROE figures per business segment.


I think your ROE refers to UK PBB with a return on tangible equity (ROTE) forecast of 32% – impressive, but just about 20% of RBSs business. What about the other segments, such as Commercial and CIB, the bulk of which is UK based? (About 1/3rd of their total exposure is ex UK according to the annual report). Returns are not that impressive, especially considering that we are still on the very nice side of the credit cycle. Why does the benefit of concentration not accrue to these segments? Is it, for example, because the international competition in these segments? I do not deny the existence of banking franchises, but the drivers of their performance have to be understood at the micro-level and such an analysis for the Greek banks I have not seen.

More importantly, and this is valid for your comment about Dexia’s core business as well: in my experience, it is very difficult to carve out the true performance of a segment of a complex banking group – much more difficult than for other businesses. Providing accurate figures is itself a very difficult task for multinational banks. Just take the RBS table above as an example and look at the horrible losses in what is labelled the “other” segment. Can you be sure that overhead costs have been properly allocated to the respective business segment? Or, could it be that the business line managers are lobbying internally to dump everything into that basket which doesn’t belong to anybody, so they can look good – politics not performance brings bonuses!

Now, it is the first time I look at RBS and I am no means an expert, but the fact that all clearly identifiable businesses perform nicely and, at the same time, the group is suffering losses doesn’t increase my confidence in their transfer price policy and their business line figures…

David Einhorn

  1.  We used 15% ROE in our estimates for a recovery. Other markets have done better. But, I won’t quibble over 14%. Should that be achieved, I don’t believe the Greek banks will trade anywhere near the current discounts to book value. for example, Komercni, the listed Czech bank, trades at 2x TBV for its 14% ROE.


Well, none of my arguments apply to Komercni: They are not in the Euro, which means if you are a foreign bank, you have additional currency risk. Plus, all non-Euro CEE countries have their own regulator and local regulations that can be very different from Basel II applicable across the Euro zone. These are all potential deterrents to would be entrants. It is much easier for the local regulator to refuse a banking license in the Czech Republic than for a regulator in the Euro zone – the ECB has no say in the Czech republic.

The macroeconomic (and indeed the historical and cultural) backdrop could not be more different from the situation in Greece:

For one, the Czechs have low government (48%) as well as low private debt (83%) which makes a (debt) growth story much more plausible. Second, they are a nation of savers which is why its loan-to-deposit ratio stands at fantastic 72% vs. 109% for Piraeus – a huge funding advantage. Last but not least: the Czech Republic is the most industrialized country in Europe, i.e. there are plenty of economically sound customers to choose from. This is not a coincidence, but a historically grown state of affairs: the southern and western part (Bohemia and Moravia) of the Czech Republic was the industrial “rust belt” of the Habsburg Empire – no such economic structures exist (or have ever existed) in Greece.

Conclusion:  the cost of capital for Czech banks is (and should) be lower than for Greek banks and its loan growth prospects better – Komercni deserves to trade at a significant premium to Piraeus. If you think Komercni’s TBV is justified, Piraeus should trade way lower (given 14% ROE).

In the next post I will respond to David’s second comment. Patience is required, as I am currently busy helping friends raise VC-money. I hope I will find time to post by the end of the week…

David EInhorn on Piraeus Bank: Now I know what he Thinks (Part II)

In this post I want to discuss some of the microeconomic, i.e. bank specific, arguments David Einhorn made during his presentation. I will only discuss Piraeus bank here, as I am already familiar with its financials due to my past blog posts. As in Part I, I will analyse his arguments one by one.

Similar to his macroeconomic analysis, Einhorn numbers improvements across all measures of bank performance and sees the potential for the share price to double over the next three years. The following slide neatly summarizes his view.


According to Einhorn, the stock price will be driven by the following three factors,

  • Cumulative “core earnings” due to increased profitability,
  • future releases of LLPs, due to over-provisioning in the past and, last but not least,
  • a re-rating of the stock: he expects the P/B multiple to double (to 1.5 x book).

Let’s have a look at each of these arguments…

Ad. 1.) Improved profitability

Einhorn has two slides that show various improving profitability indicators of which the most important one, at least to me, is the net income margin chart.


The net income margin (NI margin) of a commercial bank is comparable to the revenue figure of an industrial company, i.e. it represents the top-line. It is the most important figure by which we should judge the long-term growth potential of any company. This is because for your typical industrial company, the top-line will be unaffected by one-time measures such as cost cuts that are unlikely to lead to sustainable growth in profits – not a perfect measure but a good start.

However, banks are somewhat different as the NI margin is not only driven by growth variables (costumers, loan pricing…), but also by an (important) cost variable: the cost of funds. Indeed, Einhorn mentions that the lower cost of deposits has been the main driver of NI margin growth.

Now, by how much more can the cost of deposits fall in Greece?

In order to answer that question, I have looked up the deposit rates Piraeus bank pays to its customers. Thankfully management has provided this table in its Q2 investor presentation


Current time deposit rates in Greece stand around 2% which doesn’t sound that high to me. As we can see, time deposit rates have come down quite a bit over the past 18 months falling by more than 200 bp thereby boosting the top-line. Another such boost is impossible. The fall in funding cost has of course mirrored the fall in the Greek government bond yield over the same period, see below…


Note, that the second quarter 2014 marked the low for the 10-year Greek government bond at around 5% yield. Since then yields have increased substantially hitting 8% in November as the likelihood of another Greek default has increased. As I have argued in my previous post, given Greeks high sovereign debt burden and its high debt service costs, another restructuring is only a matter of time.

Given the already low deposit rates and the increase in Greek bond yields lately, it is safe to assume that the tailwind from lower funding cost is over. This leaves us with lending as the only driver for NI growth.

Yes, what about loan growth?

As always: facts first…

The table shows the evolution of the gross loan volumes of Piraeus bank in Greece, its most important market…


Unsurprisingly, loan volumes have been falling across all segments, despite lower interest rates. I say unsurprisingly because this is what a de-leveraging looks like.

Question: what drives this de-leveraging?

Three possible explanations:

  • Piraeus bank doesn’t have enough capital to lend to costumers,
  • it doesn’t have enough liquidity to lend to costumers, i.e. it is overstretched,
  • it doesn’t find enough good customers who want to take out a loan.

Now, it is probably a combination of all three factors, if you ask me.

Although the bank looks well capitalized on a regulatory basis, I do have a few reservations about that book value – but more about that later.

Improving the liquidity profile certainly played a role as well, as the bank’s loan-to-deposit (LTD) ratio has come down substantially. This has, no doubt, to do with bank management being eager to get rid of the ECB’s “emergency liquidity assistance” (ELA) facility, an exceptional measure by the ECB to keep the banking systems of periphery countries afloat during the height of the sovereign debt crisis. This must have had high management and local regulatory priority. After all, an ELA-free balance sheet signifies a normal environment – and better negotiating power in Brussels.

But the most important factor – and, I think, true of all EU periphery countries – is the lack of potential and creditworthy debtors. What do you expect when a high double-digit percentage of the workforce lacks (official) income? And it is not exactly the case that Greeks are unfamiliar with leverage. The table below puts Greece’s total debt load (government, households and corporate) in perspective:


Greece’s total debt load (Total ex financials, column e) at 304% of GDP is one of the highest among developed markets. Higher, for instance, than in the US (264%) and way higher than in Germany or Austria – not necessarily markets which are credited with high banking growth potential.

Actually, I think that it is more meaningful to compare Greece to other Emerging Markets. First, Greece suffers from many problems typical of emerging markets (high level of cronyism, inefficient state bureaucracy…), but also because it is EM that are seen to be “under-banked” which are usually touted as the next growth opportunity in banking circles. How does Greece  compare to your typical EM? Look at the following table for answers:


Whether the “EM-are-underbanked-story” is sound or not, at least the bankers have a point: the average debt level (private + government) in EM is exactly half the Greek level (151% vs. 304%). Note that this is not just because of Greece’s high government debt: apart from Hungary (144%) and China (169%) no other EM has as high a private debt load as Greece (129%).

Summary: the fact that funding costs are unlikely to fall further and that there is limited loan growth potential due to the already high debt levels in the Greek economy, make it unlikely that the top-line of Piraeus bank will increase significantly from current levels. If there is growth potential, I would like to know where it comes from.

Ad. 2) Release of LLPs due to overprovisioning

According to the second argument, Piraeus bank has provisioned conservatively in the past and has potential to release some of these provisions thereby increasing profits and book value. Einhorn argues that, due to a law passed in 2008 which forbade foreclosures by banks, a lot of the NPLs are “strategic” in nature and will start paying once the law expires, i.e. next year.

I am not familiar with the details of this law, but many CEE governments have passed a similar law. For instance, Hungary and Ukraine with Austrian banks complaining. Usually, these laws are passed to protect retail creditors from losing their home – not corporate customers. Now, according to the Q2 presentation EUR 19.8bn out of EUR 28bn NPL of Piraeus are defaulted business loans. If it is indeed true that said law businesses from being foreclosed, I would be very worried about it, not bullish.


Because a bank’s assets are a collection of contracts that promise a certain cash flow. A contract’s worth is a function of its enforceability. The same goes for a pledge on collateral. In countries where rule of law is weak or enforceability is otherwise uncertain, as this law seems to suggest, we do not observe complex business models emerging. In these countries it’s mostly “cash and carry” business models that are successful. Put differently, such environments are not very bullish from the perspective of a prospective bank owner. As Klaus Kastner pointed out: the World Bank singled out the difficulty in “registering property” and “enforcing contracts” as the biggest challenge to Greek competitiveness.

(Note to Keynesians out there: there is much more to competitiveness than unit labour costs reductions as a result of currency devaluations.)

In short: I have my doubts on whether these NPL will start paying again…

Now, let’s look at the provisioning level of Piraeus bank. Einhorn provides the following slide.


Just eyeballing the coverage ratios (LLPs as % of NPLs) Piraeus doesn’t seem conservatively provisioned: 16% on mortgages after a massive drop in asset prices over the past five years doesn’t strike me as conservative. Of course, if you count the collateral then everything looks fine, i.e. a coverage of NPLs larger than 100% across all segments. But it is impossible to take comfort with the collateral values, in my view, before answering the following questions:

What type of do they count here? I hope it is not property, plant and equipment. How do they arrive at the figures? Transactions? Hardly possible in a depressionary environment. Estimated values (Appraisals)? If yes, how old are they? Are the values reported nominal values or haircut values? And how much is collateral worth in an environment where enforcement is legally difficult, anyway?

Most important of all: If collateral covers the net loans, why is the NPL ratio rising and standing at a record 38% of all loans, why are they not getting rid of these loans?

The answer, of course is that the book values of the loans don’t reflect the market value of these assets – otherwise the bank would have got rid of them a long time ago. They have every incentive to do so: it looks better on the balance sheet and it frees capital and liquidity. Also do not forget that in workout speed is everything, i.e. the longer you wait to enforce collateral the less valuable it becomes. Carrying the NPLs from, say, 2010 on your balance sheet doesn’t make them worth more, on average as a defaulted customer usually will delay capex as long as possible – with negative repercussions for pledged collateral values.

Fortunately, we have another indication of Piraeus’s asset quality: the recently concluded “Asset Quality Review” (AQR) by the ECB. The following chart (via Berenberg bank) summarizes the AQR findings on a country by country basis.

The figure above shows the additional provisions uncovered by the AQR, both, on an absolute and on a relative basis. Now, David Einhorn might find this surprising, but the Greek banking system was found to have the by far biggest issues, i.e. its book values the least reliable, with Piraeus one of the biggest hit (Alpha the second bank Einhorn talks about scores somewhat better).

This confirms my view: I do not see any scope for major releases here. Certainly the ECB has looked somewhat closer at the collateral values and has not taken management figures at face value – neither should anybody else.

Of course, the regulators managed to calm the markets by saying that the capital measures undertaken by Piraeus so far suffice for now, i.e. they fulfil current regulatory ratios – nobody needs another Greek banking crisis, after all. However, I want to emphasize the adjective “current” in this context, because starting from 2019 onwards, European banks are supposed to fulfill much stricter Basel III standards. In order not to overburden them, regulators have given European banks time until 2019 to prepare for the tougher environment. The ECB knows this and consequently also tried to assess the banks’ requirements under Basel III in a forward-looking way.

Shouldn’t we as investors be forward-looking as well?

I think yes, here are the results:

This is interesting: under the fully loaded Basel III scenario Piraeus bank (line 116) ends up with a regulatory capital ratio (CET1) of  a negative (!) 1.5 percent – no doubt the doubious quality of Piraeus bank’s book value must have played a part.

I have no special information on how the ECB came up with that number, but if I interpret it correctly Piraeus bankwill have to come up with substantial equity in the next five years – it will be a long time before Piraeus can grow its loan book substantially, let alone pay dividends even if profitable.

Summary: the fact that the ECB AQR uncovered serious issues with Asset Quality in Greece and at Piraeus bank makes provision releases highly unlikely in the future. On the contrary, I would expect serious additional provisioning needs as the coverage ratios (without collateral) are not high, also compared to other banks in periphery countries. I repeat: this is no surprise, it is a well-known rule in workout management that the longer you wait the less you get back. The effect of Basel III is not even mentioned or considered by Einhorn – irresponsible.

Ad. 3) Multiple Expansion

If you follow Einhorn’s logic, golden Times are ahead for Greek banks and the economy, with growth prospects correspondingly high. He therefore goes on to assume that in three years investors will be willing to pay a premium to book value.

Apart from everything I have said so far about the dubious quality of Piraeus’s book value, it strikes me as strange that a so-called “value investor” is willing to speculate on the mood of “Mr. Market” in the future. Isn’t there supposed to be a margin of safety? If yes, where is it? This shows how far standards are  successively lowered in a bull market, even by those claiming to look for value (this happens often enough, read this Youngmoney post).

As is well-known, the P/B ratio is a function of the excess return, i.e. return over cost of capital, and the growth prospects of a company. Only for companies that can be expected to earn above their cost of capital one should be willing to pay above book value.

This relationship can be expressed with the following formula (derived from the “Gordon Growth Model“),

   P/B  = 1+ (ROE-r)/r-g)

where ROE is the return on (book) equity, (r) the cost of equity and (g) the growth rate. Note that, if (ROE = r) then the numerator is 0 and there should be no premium to book value. The nice thing about this formula: it offers us a possibility to back out Einhorn’s implicit expectations. I am interested in how much he expects this bank to earn three years down the road.

To do this, we first have to determine the appropriate value for the return investors expect when they invest in Greek banks, i.e. the cost of equity (r). I will not be overly harsh and argue that 12% is the minimum return investors demand for a Greek bank, although most investors probably would expect more in view of the macroeconomic situation.

Next, we have to settle for a long run growth rate. Given what I said about the comparatively high debt load of the Greek economy and the headwinds banks will face with Basel III, I believe it is very optimistic, if I assume a growth in profits of 5 percent per year.

Thus we get: (r-g) = (12%-5%)= 7 percent for the denominator

Plugging in a P/B of 1.5 and reshuffling terms yields an expected ROE requirement of 15,5%. This means that, in order to trade at 1.5 x book, Piraeus bank will have to earn 15.5% on book value on a sustained basis AND be able to grow by 5 percent per year – no “margin of safety” to see here.

But what does this mean in actual numbers?

Piraeus book value stood at around EUR 9.3bn at the end of the second quarter. Multiplying this figure with the expected ROE (15.5%) yields around EUR 1.5bn in expected net income (net income has been slightly positive YTD only if you include one offs) for Piraeus bank three years down the road. Note that this figure would be even 20% – 30% higher if Einhorn’s LLP releases materialize, as this would add to book value.

Is a net income of EUR 1.5bn realistic?

According to Einhorn’s slides, Piraeus bank’s pre-provision profit stood at around EUR 300 mln at the end of Q2 2014. That’s already significantly up from previous quarters due to the lower cost of funding and cost cutting measures. Annualizing this figure yields EUR 1.2bn pre provision profit, i.e. before deducting risk costs. Of course, risk costs are a real cost factor for banks even in normal years. Consequently, I assume that a bank of that size will have normalized risk costs of around EUR 100-200 mln per year. Deducting this from pre-provision profit leaves us with a normalized net income figure (we can ignore taxes due to high losses in the past) of around EUR 1 bn.

In other words: Pre-provision profits would have to increase by 50% over the next three years in order to meet Einhorn’s expectations AND risk costs would have to normalize rather quickly. This all has to be achieved taking into account:

  • no dramatic potential for further cost of funds reductions
  • limited growth prospect due to high debt load of households
  • the challenging demands posed by future Basel III requirements

And I repeat: Einhorn implicitly expects even higher net income, as he assumes book value will be higher!


This is a very weak investment thesis. The quality of the analysis is disappointingly low. Both his macro and his micro arguments are overly superficial, as he takes all figures at face value without trying to understand the drivers behind them. This makes him blindly trust the coverage ratios presented by management, or ignore the effect of known regulatory changes. He doesn’t touch the management topic at all, although there clearly are some issues such as a cosy relationship with regulators/politicians – in my experience not necessarily evidence of shareholder value driven ethics. Worst however, is the fact that he does not demand a “margin-of-safety” from this investment, but prefers to speculate on how much people might be willing to pay in the future.

David Einhorn on Pireus Bank: now I know what He thinks

I am grateful to reader IAthinker who commented on my recent Pireus bank post providing a link to David Einhorn’s presentation he gave at the recent Robin Hood conference, an annual gathering of Hedge Fund managers. At the conference, Einhorn reiterated his bullish stance on Pireus and Alpha bank, two of the biggest banks in Greece. As regular readers know, I have always wanted to understand Einhorn’s (Seth Klarman and John Paulson are long this stock too) thesis and I read the presentation with great interest.

As most presentations or writings by Einhorn they are fun to read, full of interesting charts and thoughts. He starts with the bullish case (40 slides) for Sun Edison (SUNE) – a solar company he thinks undervalued. The next 40 slides are reserved for the Greek recovery and said bank investments. He finishes by noting that France is in a mess and its bonds overvalued.

To sum it up: the solar case sounds interesting and I agree 100% with him on France. I disagree, however, with his bullish view on Pireus bank. In the following analysis I will present my view on the main bullish points made by Einhorn. Since this is a lengthy post, I have decided to divide it up: The first post analyses his macroeconomic arguments, whereas in the second post I proceed to the microeconomic, i.e. bank specific aspects of the case. For that reason I will not follow the order of the arguments in the presentation, so do not be confused.

Point Nr. 1: Improved Competitiveness of the Greek Economy

At the beginning, David Einhorn points out how the hard reform measures the Greek government has undertaken, such as slashing the minimum wage and reducing pensions, have increased competitiveness and improved business climate, a fact which has been duly mentioned in the World Bank’s  recent “Doing Business Report” (via. Klaus Kastner’s blog). According to the World Bank, Greece has improved its ranking by 40 notches since the beginning of the crisis, albeit it still ranks at the bottom among EU countries.

Einhorn underlines his argument by pointing at charts such as this one:


Now, if you ask me, the chart still shows a contraction in the actual figures with unemployment (a lagging indicator!) merely levelling off, not improving! All the positive figures are merely forecasts. And we all know that most government/IMF forecasts are – better: have to be – skewed to the positive for political reasons (politicians want to have an excuse not to undertake reforms) – in agreement with their Keynesian in-house advisors who do not want to shatter “confidence”. This, and the fact that currently all major European economies are slashing their growth forecasts, makes me somewhat sceptical.

Listen, I generally agree that the Greek government has undertaken steps in the right direction. After all, they faced what awaits all socialists at the end of the road: the end of other people’s money. This is why I have been long two Greek stocks (OPAP and Mettka) for the better part of the past two years. Under normal circumstances it might be straightforward to conclude that a more competitive economy means great prospects for the banking sector and there is the notion on Wall Street that you best “play” the economy by investing in banks. However: whoever came up with that idea (a long time ago, as it is old) did not face the current debt problem that still plagues Greek economy, as I will discuss next.

Point Nr. 2: Greece is better than France (Debt problem solved?)

Einhorn seems to think that Greece’s high debt load (public debt @ 175% of GDP) is not an issue as it is owed to supranational (ECB; IMF, Euro zone countries) at below market rates and with long tenors. He then pulls out this chart that shows Greece’s interest rate burden steadily declining and compares this to the French figure. No doubt: it suggests substantial impressive improvements ahead.


Now, the question before us: how can Greece reduce its interest burden by almost 50% without repaying anything  – it still has a budget deficit, remember – and absent any new debt restructuring?

How, David?

I got so intrigued by this chart that I even decided to look up the autumn budget forecast of the European Commission. On page 174 of the report you have the budgeted interest expenses for all Euro zone economies. I have reproduced the relevant section (Table 37) below:


And there you have it: the European Commission forecasts Greece’s interest expenditure to come in at 4.3% for 2014 and at 4.2% in 2015 – twice the figure in the presentation. I repeat: this makes intuitive sense, since absent another debt write-off there is no way Greece can reduce its debt burden by the amount indicated by Einhorn. In other words: ALL the beneficial effects of the restructurings are already in the figures – sorry, it doesn’t get better than that!

Note that Greece (together with Italy and Portugal) has still the highest public debt burden of the Euro zone, despite the default and despite the generous terms on its debt.

Note also that Einhorn got the French figure (2.3%) right – chapeau 🙂

What do we make of that?

I do not hope/assume Einhorn deliberately wanted to convey a wrong picture. It is more likely that one of his many analysts made a mistake. Nevertheless, he should have noticed that this doesn’t add-up. Worse, the mistake is not merely academic, but has real implications for Einhorn’s investment thesis as it means that the Greek sovereign still has a massive debt problem, something which has to be addressed when talking about bank investments. Unfortunately, it is something he doesn’t mention further in his presentation.

So let me drop a few words on Greek sovereign risk, then.

Obviously, the high debt load indicates a high probability of another default. Although the debt is mostly owed to foreign non-private entities, this is not necessarily an advantage. With whom do you think you can restructure your debt quicker and in a more meaningful way: with private investors, or with the EU Commission/ECB/IMF? Clear answer, if you ask me. Apart from that, there are still the constant negative repercussions of a debt overhang: political tensions and public unrest are a constant danger, higher risk perception negatively affecting bank funding and margins another and, whereas Greek households might not own much Greek sovereign debt, Greek banks surely still have high direct sovereign exposures. But more on that in the next post…

Point Nr. 3: Oligopolistic banking markets are highly profitable

After numerous restructurings and mergers during the crisis, four banks have come to dominate the Greek banking market with a combined market share (assets and deposits) of 93% among themselves. The graph compares banking sector concentrations in various countries to average ROEs implying high future profitability for Greek banks.


Now, on the surface, it is hard to argue with this thesis. Players in oligopolistic markets should earn a higher return due to limited competition. The problem at hand: what defines an oligopolistic market?

Mainstream economic theory grades markets according to the number of players operating in it: the more players, the higher the competition.

This is (mainstream) theory. What about real life?

In real life it is difficult to count the “true” number of players. You could have a lot of players in one sector, some of which occupy a niche in that market allowing them to still earn high ROEs. Alternatively, you could have only one or two players but competitive prices and average profitability. How? Well, they might be so-called “Contestable Markets” where the threat of another player entering the market keeps prices down. Airline routes are a classic example for the latter.

According to the Austrian School, a competitive market CANNOT be meaningfully defined by the number of buyers/sellers, as the classic perfect competition model suggests. What MATTERS instead is whether there is free entry/exit into a particular market. As value investors only know too well: without barriers to entry a business is poised to earn mediocre returns, no matter how many players…

How does the Greek banking market score on that goal?

As in any other country, you need a licence to open a bank in Greece suggesting the existence of a moat. On the other hand, Greece is in the EU and one of the pillars of the EU treaties is the free movement of capital. If Greek banks earn high ROEs, this can and will attract other European banks to open a subsidiary in Athens and start earning high fees themselves. It is even conceivable that foreign banks enjoy a funding advantage, due to a better reputation, as is the case for Austrian banks in some CEE countries. This isn’t going to happen in the near future, but it doesn’t look like a sustainable “moat” to me.

Wait, but what about the banks in other countries such as Belgium, Sweden and UK? Don’t they earn high ROEs?

Unfortunately David Einhorn doesn’t provide details of how he computed the average ROEs, a crucial information – especially for bank financials.

First: did he use last year figures or cyclically adjusted ones? Credit risk is a heavily left skewed distribution which means that one or a few particular observation don’t tell you much. Since bank management has more discretion about managing the numbers than in most other industries a particular annual figure is even more meaningless.

Second, if he uses some kind of historic average (better), he faces two major obstacles:

  1. How to treat defaulted banks in the sample
  2. How to deal with the fact that most of the countries mentioned have experienced tremendous increases in debt over the past 20 years

Ad. 1.)

Two of the countries (UK, Belgium) on the slide saw large-scale bank defaults during the last crisis. Dexia and Royal Bank of Scotland RBS come to mind. In both cases equity investors got wiped out. Others (like Barclays for e.g.) got bailed out, i.e. equity investors were spared,  something which I think will not be politically feasible next time. This makes historic data somewhat unreliable even if cyclically adjusted correctly.

Ad. 2.)

Sweden and Australia had no major bank defaults during the last crisis (Australia didn’t even have a crisis as it profited from China’s mad construction spending binge). Fine! But there is another thing they have in common: their private debt levels (not government debt) have soared over the past two decades. According to the Geneva Report on the World economy, the private debt levels for the Swedish and Australian Economy stand at 252% and 180% of GDP, respectively. Compared to these figures, the famously levered US private sector looks thrifty in comparison (@160%). In other words: a debt crisis and mean reversion of those high ROEs are likely to happen sooner or later. Interestingly enough, in Spain, a country which also had a much-touted low public debt load before the banking crisis hit, private debt stands at 209% of GDP – right between Australia and Sweden.

Calculating bank profitability can be tricky as I have outlined above. Almost all of Einhorn’s oligopolistic markets have experienced an Alice-in-Wonderland environment over the past two decades: high debt growth and government subsidies that have skewed historic figures. It is difficult to see how this favourable scenario will repeat for the Greek banking system (and for all others) in the future. Therefore, I feel that the substantially lower ROE (14%) of the Czech banking system is probably closer to the truth – and even that strikes me as optimistic.

In the next post, I am going to discuss bank specific statements made by Einhorn…