Paul Krugman’s false logical step

To my way of thinking Paul Krugman has finally nailed the question as to bank nationalisation that matters.  This the money quote:

That said, some decision must be reached on bank liabilities. Sweden guaranteed all of them. If forced to say, I would go the Swedish route; but of course we can’t do that unless we’re prepared to put all troubled banks in receivership. And I’m ready to be persuaded that some debts should not be honored — this is a deeply technical question.

He is absolutely right that this is the critical step in the decision making process is what parts of the banking structure you are going to either guarantee or effectively guarantee.  The critical question is not nationalisation.

Sweden could guarantee all banking liabilities because – frankly – their banks were not that deeply insolvent.

We know they were not that deeply insolvent for a few reasons – the best of which is that ex-post the Swedish bailout cost very little (and the Norwegian bailouts were actually profitable for the government).

However it is fairly easy ex-post to tell how insolvent the banking system was.  It is not very easy ex-ante to tell.  If it were easy then banks that were not at all insolvent (such as Svenska Handelsbank) would not become 20 bagger stocks quite quickly after the crisis.  The stock market would not have marked them down so much.  

The US Government’s stated position – Bernanke yesterday as well – is that there will be "No More Lehmans".  What that means is that there will be no more uncontrolled liquidations of large financial firms.  

The only way that the government can say that there will be no more Lehmans is to effectively guarantee large parts of the financial system.  That is what the statement “no more Lehmans” means.  If you want to make that statement operational you either (a) need to guarantee the banking system or (b) pour money in continuously whenever a bank (Citigroup. AIG or otherwise) threatens to become the next Lehman Brothers.

The state of US policy at the moment is nothing more sophisticated than (b) above – which is whenever an institution threatens to become Lehman the US Government tips in another 30-300 billion.  We are still in the world of the ad-hoc guarantee - of the Sunday press release.

The problem with the ad-hoc guarantee is that nobody really thinks that it is a guarantee – and the generalised wholesale run on financial institutions will continue until they are sure.  In other words we are effectively guaranteeing the liabilities without getting the policy benefit of that guarantee (which is the restoration of faith in the financial system).

Krugman has nailed the right question.  The right question is whether the correct policy is “No More Lehmans”.  I am pretty sure it is.  I think the revisionist history about how bad the Lehman failure was is simply revisionist crap.  I am convinced that at least in some instances the “no more Lehmans” policy will be operationally expensive in some instances and will leave the taxpayer with an enormous hangover*.  The alternative is simply to allow big institutions to be pulled apart by the FDIC.  Chris Whalen by contrast is convinced the other way – he says the model is easy to determine – just go down to the Southern District of New York and talk to the Lehman Trustee.  

There is a reason why the right policy might not be "No More Lehmans".  Its about cost.  If the cost of making that promise operational was $12 trillion then you probably should just let the financial system burn.  Why – because it is so much money the taxpayer could not plausibly absorb it without decades of higher taxes.  If the cost is $1 trillion then hey – just suck it up - a fast rebound to the US economy as per Sweden after its crisis is worth more than a trillion dollars.  The cost depends on the size of the banking system and the size of the losses relative to GDP.  Iceland had to let its system burn because it could not plausibly bail out its banks.  The UK banks started with very little capital and with very big balance sheets relative to GDP.  They are also problematic.  The US banks by contrast started with lots more capital and smaller balance sheets relative to US GDP.  The upper-end estimate of losses (Roubini) is $3.4 trillion.  If that is the case the upper limits to cost of the "No More Lehmans" policy is less than $3.4 trillion.

My long post has some indication of how you might estimate the costs of making a “No More Lehmans” promise operational.  I have a forthcoming post which explains quite carefully what the least cost way of making that promise operational is.   (The costs are however potentially very large - and whilst I think substantially less than the Roubini number I can't dismiss the possibility the costs could be large indeed.)  

Anyway – if you have made the decision to have “No More Lehmans” then you have made the important decision – you are going the Swedish Route and guranteeing stuff - whether by Friday evening crisis or whether by design.  I think America will go the Swedish Route – I am just waiting.  The Swedish route is guarantee and selective nationalistion.  I have never been afraid of the Nationalisation word – and anyone who buys money center banks now can expect a few of them to be nationalised.  I have small positions - which would be larger positions if I knew the rules.

But the second part of Krugman’s paragraph contains a deeply troubling false logical step.  He says: “but of course we can’t do that unless we’re prepared to put all troubled banks in receivership”.

To see why this is a false logical step you need a little history.  A long time ago most the liabilities of almost all banks were deposits.  The government guaranteed the deposits by creating the FDIC – it hence stopped crisis driven bank runs.  It increased stability in a crisis.  However it also allowed financial firms to take huge risks or even be looted (as per Charles Keating).  The solution which was adopted (and let lapse of late) was that banks got the guarantee – but were heavily regulated to protect taxpayer interests.  There was no need to nationalise the banks simply because you guaranteed the bulk of their liabilities.  There was however a requirement to (a) regulate them, (b) assess their capital and (c) take “prompt” corrective action when that capital was inadequate.  Prompt corrective action included confiscation.  You did not take over banks because they had runs (the purpose of the FDIC guarantee was to stop runs), you took over banks when they inadequate capital.**

Nowadays a lot of banks have the bulk of their assets funded by things that are not deposits.  Indeed at many banks deposits constitute less than half the balance sheet.  

The old FDIC guarantee can’t stop runs because the run that happens is wholesale – it happens outside FDIC guarantee limit.  If you want to stop bank runs the way that the original FDIC stopped bank runs you need to bite the “Swedish Bullet” – that is you need to effectively guarantee everything.

However just as the creation of the FDIC did not require you to be “prepared to put all troubled banks in receivership” a Swedish guarantee also does not require you to put all troubled banks into receivership.

What the FDIC guarantee required – and what a Swedish Guarantee will require – is you be prepared to (a) regulate banks heavily on an ongoing basis, (b) test the capital of banks, (c) force them to be adequately capitalised (rasing money if they can), and (d) nationalise the banks that cannot raise adequate capital.  

When the good times return you probably need walk away from this general guarantee.  In other words you have to regulate banks in such a way that they can’t become large enough to destroy the whole economy - so that you reduce the systemic risk at the cost of stifling "financial innovation".  That means that the recidivist Citigroup – a bank that seems to blow up every cycle – will never be allowed to become as big and nasty again.  It would be a terrible policy outcome if we did not learn from this crisis and did not regulate in such a way that it was less likely to happen again.  Willem Buiter's call for "over regulation of banks" looks right to me.

Krugman’s illogic however does not help the debate.  There is a need to guarantee all banking assets – and it should be done provided it is affordable.  There is no consequent need to nationalise the whole system – though there will be a need to have a process which will result in nationalisation of some institutions – what I call “nationalisation after due process”.   

Oh, and the number of losses in the system is not fixed.  If the ability to borrow to fund risk assets is not restored then commercial property for instance will fall until its yield becomes attractive to an unlevered buyer.  My guess is that is about 15%.  As the economy will be in a slump at the same time and rents will also fall that might mean a top to bottom move in commercial property of 80%.  If the move is that big then all the banks (good, bad, otherwise) are insolvent.  However if the banks had guaranteed funding then (a) they could lend so the slump in the economy would not be so bad and (b) people could borrow to buy commercial property so its price does not need to fall until the yield is 15%.  The top to bottom fall might be 35%.  The system losses would be smaller.

If we do not guarantee all bank funding then I am afraid that Christopher Whalen will be right - the macroeconomic wave going through the economy will just smash up everything fast.  

The longer we wait before biting the Swedish bullet the larger the system losses will be - and hence the higher the cost of biting that bullet.  Either do it now or give up saying that there will be "No More Lehmans".  If you wait too long everything becomes Lehman.  

It took Krugman a long time to realise that the "Swedish Guarantee" is the important question.  And it is.  Nationalisation (which should happen for some institutions) is only the secondary question.

John Hempton

Some post scripts

*The instances in which I think the “no more Lehmans” policy will be operationally expensive are (obviously) AIG (almost certainly) Fannie and Freddie and speculatively a few others that are properly insolvent.  My biggest problem child is Barclays – which is technically a UK institution – but it is too big for the UK to bail out – and which has a lot of its operations in the US.  I suspect that the US can – as a technical thing – let Barclays be the next Lehman – saying – hey – its not one of ours!  But that is a post for another time.  

**This is one of the things that most annoys me about Sheila Bair’s confiscation of Washington Mutual.  WaMu had a run.  The old role of the FDIC was not to make banks fail when they had runs – it was to stop runs.  I would have no objection to confiscation of WaMu if it was demonstrably insolvent.  However it was not demonstrably insolvent – and Sheila Bair’s own press release said it was capital adequate when confiscated.  It was a very strange interpretation of her role indeed that she should close a bank because it had a run.