Monday, January 19, 2009

Voodoo maths and dead banks


I am not afraid of the N word (nationalisation).  Real capitalists nationalise – meaning if the taxpayer takes the risk the taxpayer gets (any) upside.  However I want to take issue with Professor Krugman’s NYT editorial today.  Krugman accuses members of the incoming administration of believing in voodoo rituals to keep banks alive.  He takes a worthy shot at the person I most dislike amongst the continuing economic team (Sheila Bair).  But I still think Paul has his maths wrong.

Here is the key part of the article – with Gotham being a thinly disguised moniker for Citigroup.

On paper, Gotham has $2 trillion in assets and $1.9 trillion in liabilities, so that it has a net worth of $100 billion. But a substantial fraction of its assets — say, $400 billion worth — are mortgage-backed securities and other toxic waste. If the bank tried to sell these assets, it would get no more than $200 billion. 

So Gotham is a zombie bank: it’s still operating, but the reality is that it has already gone bust. Its stock isn’t totally worthless — it still has a market capitalization of $20 billion — but that value is entirely based on the hope that shareholders will be rescued by a government bailout. 

PK is wrong.  With sufficient trust Gotham is far from bust on PK’s numbers.  Suppose – and this is an understated assumption – that the normalised pre-tax spread on Gotham’s assets was two percent – say – and for the same of simplicity – the bank would earn 3% on assets and pay 1% on liabilities.

Then the bank (if it did not have the bad loan problems) would earn $60 billion on its (normal) $2 billion in assets and pay out $19 billion on its $1.9 trillion in liabilities.  The pre-tax profit of Gotham would be $41 billion dollars.

But – as Krugman suggests – the real assets of Gotham are not $2 trillion, but in fact $1.8 trillion.  The liabilities are (unfortunately) solid.  They remain worth $1.9 trillion.

Then – if the bank can continue to operate – it will earn $54 billion on its assets still pay out $19 billion on its liabilities.  Pre-tax profits will still be $35 billion.

If the bank runs for three years it will again be solvent.  If it runs for less than six years it will be fully and adequately capitalised.  This is in fact how the Japanese mega-banks recapitalised.  I blogged about it here.  At the spreads in America – which are several percent – the recapitalisation will happen much quicker than this and much quicker than in Japan.  Indeed it is likely that with quasi government guarantees for bank funding and market rates for bank loans the spreads would be over five percent in America right now.

Paul thinks the bank has value only because there is a perception that it will be bailed out.  I think it has value because it still has positive operating cash flow (provided it does not have a run).  The value - which I believe is large - might mean that widespread nationalisation is (ex-post) profitable for government - though it may not be profitable on an (ex ante) risk adjusted basis.  

Paul Krugman might consider it voodoo economics to give implicit guarantees to banks – but the Japanese experience shows – and this post explains – that things that stop a run will eventually recapitalise a bank.  It worked in Japan.  It was not a particularly pretty way to run things from a macroeconomic perspective – though people like Nihon Cassandra think that Japanese capitalism works pretty well.  

Paul is accusing Sheila Bair et al of voodoo economics.  I am inclined to agree with almost anything nasty about Sheila Bair.  But in this case PK is publishing voodoo maths.  



John Hempton

23 comments:

Amnon Portugaly said...

A very interesting analysis. One can draw a conclusion that there is no need for government or the Fed to invest these huge sums of public money in order to bail/save the banks.

John Hempton said...

Not sure. There is CLEARLY a run on bank funding. Almost no bank in America can obtain senior unsecured debt.

Most have plenty to roll. When they can not roll it they fail.

LIQUIDITY not SOLVENCY is the core issue.

The government has made many moves that have exacerbated this. I will go back to them in another post.

J

CouldMarxBeRight said...

Why do you think wholesale funding is an issue for the US banks? Isn't there a TLGP that is working pretty well for them? The FDIC guaranteed issuance is robust and it can go ok to replace 125% of their debt outstanding. And so I have thought that the isuue with shitti group is more of a dearth of tangible common equity than about liquidity.

John Haskell said...

John- Krugman got nothing wrong. You conjure up this hypothesis that C could have deposit funding as stable as "77" Bank. It can't.

C still has market value only because of the prospect of a government bailout. You quibble with Krugman's definition of a bailout. Confined to the op-ed format Krugman only discusses the possibility of a purchase of C's assets at a fictitious price.

Not having such restrictions on space you point out that C can also be bailed out through government guarantees on its liabilities such that the American bank customer is screwed through an artificially high spread being offered to C.

As an American taxpayer and unfortunately also a user of US based banks' services I agree with Krugman and am not interested in the USG pursuing the strategy you outline.

Richard Smith said...

Nice analysis.

Was going to ask about the funding gap issues but here you are in the comments tackling that one.

There is also the off balance-sheet stuff: e.g. with Citi there is reputedly another 1.2Tn off balance sheet, with half of the asset side of that in MBS (dunno what sort, alas).

This, I suppose, extends the wait time for Citi's recovery beyond 5 yrs; unless, I suppose, the level of default is so horrible that it kills the cash flow. It comes down to what the net interest spreads actually are.

John Haskell said...

Do you think you could help your readership understand better the consequences of the course of action you propose by outlining which currently bankrupt US companies could come out of bankruptcy if given an explicit government guarantee on all liabilities so they could borrow in the 90 day market at 0.25%?

The more I think about it the more I like it. Bankruptcy would become completely unknown within the 50 states! Imagine the stimulus this would provide to America's fabled entrepreneurial class! Even Bernie Madoff would have been spared his recent discomfiture.

Of course, if you had in mind that this marvelous gift be given only to bankers, that would be different. Reading your post it seems that not only should this gift be limited to bankers, it should be preferentially bestowed on those bankers in direct proportion to their losses due to incompetence during the recent past.

There remains one mechanical problem. The only country in which this program has been successfully implemented has only one political party. The US alas has two. The non-implementing party (call them the "Donkeys") might have an electoral edge over the implementing party, which we might call the "Bushes."

Although who knows. With a good enough PR machine you might even get the electorate to believe that the Bushes "exercised bold leadership to solve America's problems."

Certainly we can expect the bank beneficiaries to give generously to the election campaign of anyone who undertakes the program you outline.

Steve said...

The analysis makes assumption upon assumption. It is speculation at best. Although I did not check it out, I have no doubt the math is correct! Steve

Boonton said...

I posted this question PK's NY Times blog but it is awaiting moderation so I'll ask it here as well.

Let's say the gov't borrows $300B at 3% for 5 years. It then deposits the $300B into Gotham Bank in the form of a 5 year CD paying 4%. Assume that Gotham is able to make consumer/business loans at a rate of 6%, after costs, so in essence it has a 2% profit margin.

1. In terms of liquidity this helps Gotham since it knows that the gov't will not withdraw the CD until 5 years from now.

2. In terms of lending, Gotham has a strong incentive to lend since it must pay the gov't 4% and treasury bonds will only yield 3%.

3. 5 years from now, Gotham will have cumulative profits $30B ($300B * 2% * 5 years, I'm ignoring compound interest for simplicity). This can partially offset write downs from the $400B in suspect assets.

4. 5 years from now, Gotham & the market in general should have a much clearer picture of exactly how much those $400B in mortgage backed securities are really worth.

5. Hopefully 5 years from now the economy and financial system will be stronger so Gotham can either find new funding from the market or if it has to liquidate itself in bankruptcy the process can be done with less choas than would be experienced today.

6. The taxpayer does not have to worry about trying to collect on $400B of bad mortgages. He basically lends money out at 4% that costs him only 3% thereby having minimial risk.

Since by #6 this is essentially costless to the taxpayer you can make the CD as large as you want. If $300B won't do the trick then make it $400 or even $500. At this point Gotham shouldn't need to roll over any large amount of loans nor should it be vulnerable to any serious runs. (If it does the FDIC & Fed would come into play but if lots of regular depositors suddenly take their money out some other bank would have to get the bulk of that).

Shareholders would be hurt to the degree that those $400B in toxic assets are worth less than $400B but not so much that the bank has to collapse.

Now for Extra Credit:

Simply expand the program to all banks. The gov't borrows a huge amount at 3% and puts it in CD's of numerous banks, large and small, for 4%. The banks that are not like Gotham, that have good assets, will be all the more liquid and richer. Gotham like banks will be helped but not as much as their competitors. Should Gotham fail other banks will be better able to pick up the pieces with minimial taxpayer risk.

Any thoughts guys? Would this work?

dWj said...

Haskell makes a reasonable point without resorting to subtlety. It's the case, though, that any financial institution, no matter how well-run, will fail if it loses access to funding. (This is even true of many non-financial institutions, but not nearly as forcefully.) If an ultimately healthy institution retains access to credit, it will ultimately be able to pay back all its creditors; if you provide funding to an unhealthy institution, you will ultimately find yourself having to provide more and more funding, without ever being able to get money back out. The question here is which of these institutions have a benign equilibrium, and which ones are bound to fail. I've tended to support some of the bailouts on the theory that the worst of the financial companies have already left the gene pool, and that the remaining ones tend to have more good than bad left in them. I'm not at all certain of this, and Citigroup is one of the places where my certainty is particularly low.

Anonymous said...

Except that in the case of US banks, their former profit model is no longer fully functional. They can't broker significant numbers of loans for resale overseas, they can no longer offer insurance they can't pay out on (futures), and fees for credit cards et al are dropping (because they are paring credit way back).

Japan had a large reservoir of stable domestic savings to work with. Japanese banks could afford to take many years to slowly figure out a new business model. Their deposits weren't going anywhere. The US bank model was dependent upon volatile foreign savings, which are now gone with the wind. Financial sector profits are now so low that it could take a very long time for profits to make them whole.

The experiment of running an entire, large economy on foreign credit is pretty much a new experience in human history. This means it is not possible to look to other nations' experiences to figure out what will happen. Other nations use a different banking model than the US does. Other nations' banks mostly loan out domestic savings (their citizens are inclined to buy capital goods with their savings).

sugam said...

John, I think your argument is not entirely correct as well. The cash flow from the assets could go down as well, a simple example would be deliquent mortgages. The asset ( the mortgage ) used to give 6% interest but those cash flows have dried up.. that reduces the profit of the banks. The Assets have fallen in the anticipation that the yield should be higher since there is a reasonable chance that more of the mortgages could go deliquent or even foreclosed.. in which case the loss on the asset side has to be realised. Note in both deliquency and foreclosure no healing happens due to those assets... could you point out if I have overlooked some aspect?

Carlomagno said...

"Then the bank (if it did not have the bad loan problems) ..."

Cough, cough... And if the moon were made out of green cheese...

Is this also how you arrive at the conclusion that liquidity not solvency is the core issue?

dansweet said...

Great post. I appreciate that you take the time to put forth real ideas with real numbers and assumptions attached to them. You avoid resorting to hand-waving and use numbers with more than one significant digit. I wish more people would have the courage to do the same. This a great example of why I keep reading your work.

cap vandal said...

The analysis - if you want to call it that - is essentially correct.

If you look at the 1930's, half the banks didn't fail. General price levels dropped over 30%. Any bank with under 30% capital ratios had to be insolvent. The fact that they did survive indicates that they don't need to be solvent in the sense that commentators seem to believe.

Part of the issue is that the money centers absolutely did themselves in with securitization and putting themselves in a position where they were required to use derivative accounting on a large portion of their assets.

I am not encouraged that they managed to lose their capital BEFORE the country experienced a deep recession. They still have to deal with loan issues on consumer loans (credit cards, auto), commercial real estate, general business loans, etc.

The real problem is that the bank model requires some degree of imbalance in asset/liability duration and they inherently have a liquidity management problem.

I personally favor hiving off all asset backed securities and simply letting them amortize. There is more then one way to accomplish this, but the idea of forcing banks to liquidate in the midst of the credit market from hell isn't it.

By the way, if you think about banking from this perspective, you realize that it is the ENTIRE balance sheet that needs to be considered from an economic as well as a strictly accounting basis. That is, if interest costs on demand deposits falls (the current situation), the liability side of the balance sheet shrinks also. Not to mention the rather idiotic but necessary consideration that the bank's preferred stocks have fallen in market value and hence, on a m2m basis, have improved the banks capital position. This has already been recognized to the tune of a few billion, but is so far limited in scope. However, it is totally consistent with derivative accounting concepts.

KLM said...

John Haskell:

Unless you are arguing that Citi is infinite-time insolvent (or actually, even then), your argument seems to be based entirely on ideology. The situation is as easy to grasp as it is to solve; the banks' creditors are not convinced they will get full value on the dollars they lent, and would rather hoard than lend: a credit squeeze is born. The government then needs to do the job it is supposed to do, namely instil confidence in the market that most investments are indeed safe, and that investing is better than hoarding.

Solution: Make sure it does its damn job as a lender of last resort! Citi cannot fail. Inverstors all over would go into röthschreck. Credit would be non existent. Bad. Is it fair to all the people who did no worse than Citi and does not get a bailout? No. Can it create a "heads; I win, tails; you bail me out" mentality within banks? Probably - if done the wrong way. And there are many wrong ways. But one that does work, and has been proven to work, is complete nationalization.

Think about it: The government will take over all liabilities but also all assets - if there is something to salvage, at least taxpayers get the good parts as well as the bad. If they are indeed truly and utterly bankrupt, think of it as the cost of avoiding a recession becoming a depression. If not, governments can (and has) end up making a profit in the long run. By a complete takeover, the government also gets to control the spending and the loss may be capped.

And you also punish the stockholders (they loose their investment) so that they will not repeat the mistake (yeah, fat chance...), even if the lenders are let off easy.

Or we could just sit still, say our "Hail Reagan" every night, and hope for the best.

Bill Luby said...

John,

I hope it is not too late to offer up my thanks for an excellent article and a consistently compelling blog.

While I have subscribed to your feed for a long time, I used this article as an excuse to link to your work and add your site to my blogroll, even though a number of other insights were probably at least as deserving.

Thanks for consistently giving me ideas to chew on.

Cheers,

-Bill

jimsum said...

This proposal of letting banks operate until they make a profit reminds me of the sure-fire strategy for winning at blackjack in Las Vegas. You bet $1; if you win great, if you lose you double your bet. Since you can't lose forever, you are guaranteed to make money with this scheme.

The usual argument against this strategy is that is possible to lose enough times in a row to wipe out any bankroll. Fortunately, this argument doesn't apply to the government's infinite bankroll, so by all means let's keep doubling down on those banks until they inevitably succeed.

MattZN said...

Generally speaking why would the U.S. government want to take over a bank that is still able to service its debt and still able to perform all the functions a bank should perform, including originating new loans? I'm a little confused why at least one prior poster believes that banks have stopped making loans when they clearly have not stopped.

The other issue is simply... what would the U.S. government do with a big bank? They certainly can't run it any better. They can't clear the books without wiping shareholders or, if trying to clear the derivatives portfolio, potentially causing a systemic failure and chain reaction. The FDIC does not have the cash anyway and the government would certainly not recover the entire deposit base from the loan portfolio (because those are 30 YEAR LOANS, not because they are worth less). It would probably take a decade or longer to recover a reasonable return from the loans.

So many people are screaming nationalization today and don't seem to understand that nationalization will not accomplish what they think it will.

-Matt

Anonymous said...

This argument is wrong. Essentially you are saying that the income stream from the assets is greater than the income stream from the liabilities - then how can the value of the assets be less than the value of the liabilities (as you assume)? It must be that the discount rate is higher for the assets. Your analysis ignores this. And there is a very good reason that the discount rate is higher - the assets are riskier. Therefore there is a significant chance that the income from the assets will be different from its expected value - ie by ignoring risk, your analysis ignores what is perhaps at the core of the crisis - an increase in the risk (default risk / liquidity risk /...)

Pat Gillett said...

John,

Great blog, great post, couple of questions...

- If a government guarantee restores trust in banking, wouldn't (shouldn't?) that reduce the spreads which banks are currently able to charge on loans - thereby extending - perhaps indefinitely - the recapitalisation period?

- If so, should the banks be allowed to keep spreads sufficiently (and artificially) high 3-5 years to recapitalise?

JKH said...

Citibank is the combination of a zombie asset bank and a profitable income bank.

Separately, the zombie bank operates on a 1 day time horizon via mark-to-market. The income bank operates on a time to maturity horizon via accrual accounting.

The zombie bank sabotages the entire bank by accelerating recognition of zombie asset losses over a time horizon that is completely inappropriate in the context of a viable income bank.

The combination would be viable if accountants would simply understand the difference between disclosure and capital charges, and let the income bank absorb eventual actual asset losses over a sensible matched time horizon.

Accountants are blinded by their desperation for transparency. There are other ways to get transparency than by treating commercial bank balance sheets like they were day trading operations.

JKH said...

The financial crisis is neither a liquidity crisis nor a solvency crisis.

It's a crisis in the definition of solvency.

Ben G said...

"If the bank runs for three years it will again be solvent. If it runs for less than six years it will be fully and adequately capitalised."

I think you're agreeing with Krugman. Gotham is insolvent. Where you disagree is in what that means. You say it's worth a shot to see if the bank can become solvent again on its own. He disagrees.

General disclaimer

The content contained in this blog represents the opinions of Mr. Hempton. Mr. Hempton may hold either long or short positions in securities of various companies discussed in the blog based upon Mr. Hempton's recommendations. The commentary in this blog in no way constitutes a solicitation of business or investment advice. In fact, it should not be relied upon in making investment decisions, ever. It is intended solely for the entertainment of the reader, and the author.  In particular this blog is not directed for investment purposes at US Persons.