« Remembering prisoners of war | Main | More evidence on minimum wages, employment and poverty »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

The solution to this dilemma is to divide troubled banks into "good bank" and "bad bank" subsidiaries. The bad bank picks up all of the legacy loans and legacy funding. The good bank can then make new loans, and get new funding, at the 5% level.

Why haven't they tried this in Ireland?

Dave: That sounds right to me. I don't know why they haven't.

Ireland is in the EU and would have to clear that with the European Commission. Rescuing a dying bank by excising its cancerous bits can be seen as contravening state aid rules. Once you remove the bad loans, what do you then do with the deposits that those loans backed? You'd need extra state aid which Ireland may not be able to provide, plus you'd have to restructure the bank's deposit book which could be difficult.

You leave out an important scenario: in a zero interest rate environment, no bank is bad. This is why otherwise insolvent banks like Bank of America or Citi can stay solvent. It doesn't matter the proportion of non-performing loans on the asset side as long as its cost of funds is minimal. Banks are thus engaged in a race with time to capture a positive return to recapitalize before interest rates rise.

The other important element not mentioned is reserve banking: commercial banks can (in Ireland too) create new money which they can then loan out.

The current reserve rate is somewhere around 10% in the eurozone, so for every 1 euro of desposits (people willing to loan money to the bank) the banks are allowed to create up to 10 euros they can lend out to others. (They are allowed to create money 'out of thin air'.)

So if they can lend out twice the amount of hard deposits, for 5%, they can already get 2x 5% == a 10% leveraged return rate - higher than the 8% cost of financing.

Of course they can only loan out so much - and with a critically depressed Irish economy (made worse by mind-blowingly stupid 'austerity' measures) it becomes harder and harder to find a good home for that leveraged money. The local market is quickly saturated so they try their luck on the global markets - just like their US counterparts.

All in one, they are in a liquidity trap: plenty of liquidity and reserve interest rates are near zero, but still there is a big gap in demand that generates a lethal spiral of contraction. That spiral devalues their other assets and causes even more defaults by individuals and corporations.

No happy ending is in sight as not even the ECB seems to understand that the first few months of the Great Depression is being replayed here, in slow motion.

Nick, there is an old European solution to the problem you describe. It is called covered bonds. Banks are able to issue bonds that have lower credit risk as they are collateralized by a specific assets. German covered bonds (Pfandbriefe) have an old history that reaches back to 1769. This is a better solution than securitization as covered bonds are guaranteed both by the issuing bank and by specific on-balance sheet assets, so principal-agent problems that have killed the securitization markets do not apply to covered bonds.

The good-bad bank solution was tried in Ireland. Government has recently sponsored NAMA, that has purchased all the "bad" assets from the banks, so all those banks you are reading about are "good" banks.

The real problem that these both two solutions (covered bonds and good-bad bank split) no longer work in Ireland. Covered bonds sometimes trade lower than corresponding regular bonds (another violation of EMH arbitrage), and "good" banks are indistinguishable from "bad" NAMA. The solution is to split Ireland in two (three if you include Northern Ireland), so the "good" Ministry of Finance can function and serve the export sector after the "bad" Ministry of Finance that serves the housing sector collapses :)

My understanding of reserves is that if the reserve is 10% then the bank can lend only upto 90% of their deposits. I dont think they create money out of thin air.

Why can't you start a new bank? Or is the risk for Irish banks not that their loans are underperforming and they are undercapitalized, but that their potential new loans are too risky for the rate of interest they can charge, given that the housing market could collapse because of bad bank insolvency?

Lets assume we start with $1000 of hard cash deposits, put into bank A.

Bank A can loan out $900 of it, keeps $100 as reserves.

That loaned out $900 is 'new money', and is present in the customer's account who took out the loan.

That account balance is 'hard cash' he can use to pay for services. It can be (and, since it's a loan, most likely will be) transfered to someone else.

That someone else has that $900 in bank B, which keeps $90 as reserves and can loan out $890 to others. $890 in 'new money' is created. That 'new money' is not treated separately - it's cash and transparently transferred between economic actors.

If we treat all Irish banks as a single entity, and ignore foreign trade for a moment, we see this banking system loaning to itself, and multiplying up the monetary base - with a relatively high leverage factor. A single initial deposit of $1000 allows a lot of 'new money' to be created, with a high multiplier.

"The solution to this dilemma is to divide troubled banks into "good bank" and "bad bank" subsidiaries. The bad bank picks up all of the legacy loans and legacy funding. The good bank can then make new loans, and get new funding, at the 5% level."

Couldn't you achieve substantially the same result by subordinating all the existing bank borrowing to new bank borrowing (or, and it amounts to the same thing, capitalizing existing debt and issuing new debt). Essentially, the new lenders would get first dibs at the new (and, hopefully, good) loans (i.e., assets) of the bank and wouldn't be exposed (in theory) to the bad assets. In theory, the bank should be able to borrow the "new" money at the lower 5% rate (assuming that the ultimate bank loans were "good" loans) because the new lenders would have recourse not only to the new "good" loans, but also any residual value from the old "bad" loans before the old lenders get paid. And, assuming the bank can actually earn a spread on its lending, the arrangement might be beneficial for the old lenders as well, as the retained earnings would increase the pool of funds available to repay the "old" borrowings and maintain the bank as a going concern until the quality of the "old" loans improve (i.e., until property prices rise).

In practice, I think Andrew F is probably right that the problem is not solely the existence of bad loans on the books, but also the considerable uncertainty about the status of new loans as "good" loans worthy of the 5% interest rate.

Hi Nick,

My sentiment is that the bank will borrow at 8% and lend at 5% because the bank is a factory. Their marginal cost is presently higher than their marginal revenue, but they continue to over-produce because the short-term loss is calculated to be less expensive over the long-run than the alternative... Charging higher rates would threaten confidence in the bank, put a hold on lending, take away existing customers (though default), and likely kill the bank's long-term viability. The bank knows how bad this option is and will choose to operate at a loss until they run out of money, hoping that the the situation improves before then.

To improve the bank's situation, one must address the core problem of the unreasonable mortgage contracts. These which can be transfered (ie. taxpayer subsidy), re-priced (ie. inflation) or cancelled (ie. default). It becomes an allocation challenge to determine who bears the cost. Raising inflation is not an option in thse case, so the government backs the bank and you end up with taxpayer-subsidized home ownership.

I don't believe that the tragedy of the commons applies to the ultimate lender: this lender profits directly from their investment and needs no futher involvement with the bank. Instead, I think you could argue that it applies to the taxpayer subsidy, in that if not enough taxpayers vote not subsidize their bank, they all suffer.

I find the Morgan Kelly article interesting because it shows the social negotiation taking place over how much the taxpayers are going to pay. I suspect they'll end up paying the bulk of the cost, allowing a handfull of defaults to trickle through, though I wonder what the long-term social effects will be? It's difficult for the individual taxpayer to see how s/he benefited. Will a backlash push the Irish into being more conservative about other issues?

"My sentiment is that the bank will borrow at 8% and lend at 5% because the bank is a factory. Their marginal cost is presently higher than their marginal revenue, but they continue to over-produce because the short-term loss is calculated to be less expensive over the long-run than the alternative... Charging higher rates would threaten confidence in the bank, put a hold on lending, take away existing customers (though default), and likely kill the bank's long-term viability. The bank knows how bad this option is and will choose to operate at a loss until they run out of money, hoping that the the situation improves before then."

That would make sense from the Bank's perspective, but why would new lenders want to be involved in such a bank at 8% or any other interest rate. Who lends money to a business with a strategy guaranteed to lose money? I suppose it might make sense if you think that the "badness" of the underlying bank assets is a temporary phenomenon likely to be reversed in short order. This, after all, was the thinking of the people who lent billions to GM over the years on the theory that, yes, the Japanese were beating the stuffing out of it, but hey, they'll turn around.

Then again, how did that work out for GM's creditors (turned shareholders)? If the "badness" of the underlying assets is a structural problem, and lenders know it, they're not likely to be incline to lend to a bank at 8% (or any other rate) if the bank can't use the money profitably. The problem facing the Irish banks is that they lent a ton of money to purchase assets (real estate) at inflated prices. That isn't a problem that's going to go away in a few years (it's been 20 years, and Japan's still recovering from its property bust of the late 1980's).

The only way out of this is for the banks to work out a solution that shelters new investors from exposure to the risks associated with the existing "bad" loans (the "externality" that Nick identifies). Different capitalization structures, covered bonds, subsidies, "good" banks and "bad" banks are all different ways of doing that. Otherwise, there'll be a disconect between the cost of capital for the banks (8%) and the price at which "good" borrowers can lend (5% assuming that there are potential retail lenders without exposure to the "bad" assets who are willing to lend in that market - say Canadian banks who are busy snaffling up subsidiaries around the world).

That someone else has that $900 in bank B, which keeps $90 as reserves and can loan out $890 to others. $890 in 'new money' is created. That 'new money' is not treated separately - it's cash and transparently transferred between economic actors.

Correction: $810 in 'new money' is created and potentially passed along to a new debtor. (Who can again spend that money, which will show up as a deposit somewhere else.)

So new money is created 'out of thin air' in a recursive way, and the total of new money can be as high as 0.9+0.9*0.9+0.9*0.9*0.9+... - which is a series that is not bound - so in theory an infinite amount of new money can be created in the reserve banking system.

In practice there's a limit of how much 'debt money' will be used to create new debt money - but the ultimate limit is determined by the banks themselves, not by any regulatory measure like the 90% limit that was mentioned. The system has much more debt than it has deposits.

This all works fine as long as there's a healthy expansion of the economy, debt is rarely defaulted upon and everyone in the chain pays their dues.

But, in days like these, when there's an excessive demand to pay back debt, economic output falls (which corresponds directly to the ability to pay back), and the rate of default increases. Banks are exposed to the rate of default in a debt-proportional, leveraged way - with no easy way back to deleverage.

Basically, the contraction in GDP falls almost straight on the banks. Foreclosure does not solve anything (even in countries that do not engage in robo-signing), as it just pushes the missing debt-money from the home-owner to the bank. (the bank wont be able to sell the house either.)

Reserve banking coupled with the inability to increase inflation (that is the case for Ireland, which does not print its own money anymore), pretty much has a deflationary death spiral built into it.

And we know how far such a death spiral can go: Latvia with 10 quarters of recession, higher than 20% unemployment and -25% GDP from its former peak. And that without any banking sector excesses (Latvia has virtually none), just the pure deflationary effects of the eurozone ...

Ireland does not seem to be able to 'earn' its way out of that situation ... (let alone Greece)

White Rabbit: You are computing the money multiplier effect of injecting $1000 of new equity into the banking sector.  And the series converges.  In your example the sum is 9.  But that's not the effect of a one banks decision to make a $1000 loan.  The effect of that is to increase the money supply by $1000. Period.  

But what if the bank has 3/8 of the market share for deposits, and deposits are yielding 1% (M2) or 0% (M1).  Then the bank funds 3/8 of the loan essentially for free and the effective cost of funding is about 5/8 * 8% = 5%. For banks that are very large in their market you can't ignore the effect of increased deposits in the marginal cost of lending.

Bob: The risk to lenders is priced in by the 8% interest rate. Any new lenders clearly hope that the risk is properly priced. If the bank's balance sheet gets worse then the rate ought to increase (though, like GM, that doesn't always happen). This would obviously be a disaster for the bank.

I agree that restructuring the bank's assets is valuable as it clarifies the risk, and enables the bank to attract investors. However I don't think that restructuring is enough to cure the underlying problem of crummy mortgages. Someone has to take on the cost of the bad mortgages, which will demand a higher interest rate than 8%.

The bank won't pass this cost onto the bad mortgage holders (as this would force them into default), so the government intervenes with a subsidy, holding mortgage rates at 5% for everyone, injecting billions into the real estate market and elevating prices. The homeowners are still paying the full cost for their original crummy mortgages, though now part of it is to the government tax collector, rather than the bank. Most people stay out of default, believing it's through their own actions, and that someone else's mortgage was the problem.

I imagine all of this will be a bitter pill to swallow..... Presumably the recovery time would depend on how overpriced the real estate is. If real estate is 20% overpriced, and inflation is around 2%, then nominal house prices would need to be flat for a decade. Meanwhile, income tax will have to go up, and Ireland isn't really a country known for it's political stability.

As far as I can figure it out, the proposed solutions: good/bad bank; subordinating existing debt; covered bonds; are basically the same. Or rather, different ways of doing essentially the same thing, which means breaking the pool into two.

The abstract theorist in me, though, says there must be something wrong with all those solutions. The whole point of a bank is to pool. And when you break the pool up, it's not as big a pool, and the risks are no longer as diversified.

But then, if that were the whole truth, there would ever only be One Big Bank, to diversify as much as possible. No new bank could ever start.

But it might all depend on the correlation between the old assets and the new assets?

And then my brain fails.

Nick, covered bonds preserve almost all benefits of pooling on the asset side of the bank while greatly reducing the externality on the liability side of the bank. Yet covered bonds no longer work. Because the problem is externality on the level of sovereign debt.

K wrote:

White Rabbit: You are computing the money multiplier effect of injecting $1000 of new equity into the banking sector. And the series converges. In your example the sum is 9.
Indeed you are right that it converges to 9 and is not infinite. I.e. basically a reserve requirement of 0.1 can multiply money up to 1/0.1, i.e. 10x.

My main point remains: it's certainly much larger than the initial $1000 - the other $9000 is created 'out of thin air'. It also means that defaults in equity (or reduction in price of equity) strike back with a factor of up to 10x, because when $1000 equity is removed from the system, the up to 10x chain of 'debt money', which was based on this original equity, has to be removed as well, to meet reserve requirements.

I do not claim reserve banking is inherently bad: it is certainly a tool proven useful when an economy is growing, it auto-scales the monetary base with economic activity and induces economic actors to not just sit on cash and seek rent, but to involve in economic activities, take risks and such.

But it certainly does not seem to be designed for stability in face of deflation with a strong external currency (such as the Euro in Ireland): the unrolling of debt can be a catastrophic, near exponential event. And in Ireland debt is apparently being unrolled on a massive scale.

But that's not the effect of a one banks decision to make a $1000 loan. The effect of that is to increase the money supply by $1000. Period.

Are you sure that it needs multiple banks to create new money? What if there's some external agent with an 'innovative' instrument which creates a 'loan' and then a 'deposit' (secured by that instrument) back to the same bank, allowing say two hops of debt and deposit, creating $2000 in new money that can be spent, based on only $1000 of original equities?

I.e. are banks really limited in their ability to print new money with up to 10x leverage - especially when they are in a desperate economic situation?

But what if the bank has 3/8 of the market share for deposits, and deposits are yielding 1% (M2) or 0% (M1). Then the bank funds 3/8 of the loan essentially for free and the effective cost of funding is about 5/8 * 8% = 5%. For banks that are very large in their market you can't ignore the effect of increased deposits in the marginal cost of lending.

Is this ever a realistic scenario in Europe, where reserve requirements are handled by the ECB so the 'market for deposits' is the total eurozone?

White Rabbit:

I think I was unclear.  The effect of injecting $1000 of equity into the banking sector at a constant leverage ratio of 10 is $10K of new money (in reality somewhat less - maybe around $7-8K since banks use other forms of debt financing as well as money).  But the scenario in which a bank decides to increase its leverage a bit by creating a new $1000 loan is very different.  If the other banks haven't changed their leverage or equity, then all that's happened is the creation of the loan and $1000 of new money.

"Are you sure that it needs multiple banks to create new money?"  No.  One bank; one $1000 loan; $1000 of new money.

"Is this ever a realistic scenario in Europe?"  Not sure.  I don't think the multiplier filters out evenly over the currency zone.  If a bank is dominant in a region and increases lending activities, then it will be stimulative to the economy of that region and a disproportionate amount of the money will stay there. So I think it has more to do with dominance in an economic zone rather than currency zone.

Nick said: "As far as I can figure it out, the proposed solutions: good/bad bank; subordinating existing debt; covered bonds; are basically the same. Or rather, different ways of doing essentially the same thing, which means breaking the pool into two.

The abstract theorist in me, though, says there must be something wrong with all those solutions. The whole point of a bank is to pool. And when you break the pool up, it's not as big a pool, and the risks are no longer as diversified."

I agree with you insofar as the point of a bank is to pool UNCERTAIN risks. The problem here is that, for new lenders, the risks associated with the existing loans are not uncertain - they know (or at least, they believe on reasonable grounds) that those loans are riskier than any new loans. In the absence of some mechanism to segregate exposure to the old "bad" loans and the new "good" loans, what you're pooling is uncertainty with respect to new "good" loans with certainty with respect to the "bad" ones. For obvious reasons, that's not a tempting bargain for new bank lenders, at least not at a interest rate commensurate with the risk associated with the new loans.

I suppose the other way of thinking about it is that the "bad" loans were not priced like "bad" loans. So, for example, you might have a bank that makes both "good" loans and "bad" loans (say insured mortgages and subprime mortgages). Such a bank should be able to raise money at an interest rate that reflects the asggregate risks associated with that portfolio of loans and still make money by charging appropriate interest rates to the "good" and "bad" borrowers. The problem in Ireland (and elsewhere) is that the "bad" loans were originally priced as "good" loans. So an Irish bank that borrows based on the level of risk of its whole portfolio of loans to make "new" good loans will lose money.


Scott said: "Bob: The risk to lenders is priced in by the 8% interest rate. Any new lenders clearly hope that the risk is properly priced. If the bank's balance sheet gets worse then the rate ought to increase (though, like GM, that doesn't always happen). This would obviously be a disaster for the bank."

Is the risk priced in the 8% (or any other) interest rate? It's worth noting that the Kelly article suggests that private sector lenders aren't willing to lend to Irish banks at any interest rate. On his account, the 55 billion Euro worth of Irish bank bonds that matured in September were only repaid by borrowing from the ECB and the ECB has intervened on the inter-bank market to bail out the Irish banks - in order to maintain the solvency of the bank's lenders - and the banks are only going to survive in the long-run on the backs of taxpayer funded bail-outs. I think that's more consistent with my story that lenders (or at least lenders who respond to economic, rather than political, forces) aren't going to lend to money-losing banks at any interest rate that the banks will (can?) pay.

Maybe they know it's a stag hunt and they've decided to hunt stag.  If the banks are all in trouble, but they all decide to lend at 5% in proportion to their balance sheets, then they will all increase their deposits by an amount equal to their new lending.  That will be profitable.

K,

I suppose that would make sense, although we'd have to ask ourselves why people would continue to deposit their funds with potentially (or actually) insolvent banks. I'd suggest that answer is the explicit guarantee of bank deposits made by the Irish government. As I noted above, government subsidies are one of the mechanisms that banks can use to insulate "new" lenders from "old" bad debt.

On further thought, the other way that lending at 5% would make sense is if you're lending money to people who buy property that's subject to existing mortgages (and, presumably, underwater). In that case, the bank doesn't have to borrow new money to finance the new loans since, presumably, the money that's lent out will come back in as part repayment of the "bad" loan (this only works if the new loans and old loans are internal to the bank or if there is coordination between banks - your stag hunt story). In this case, the bank isn't really changing it's balance sheet, it's just changing the identity of its borrowers (well, to an extent, until the old debt is written off, presumably the old borrower is on the hook for the balance). In that story, though, the bank isn't borrowing money at 8% and lending at 5%. It's taking money that it already borrowed at 5% before the bubble burst (obviously, this ignores the injections of cash from the ECB and the Irish government - i.e., government subsidies) and re-lending a part of it at 5%.

Nick,

I'm not sure that your paradox is specific to banks. It would apply to any distressed business. For example, how would a distressed company with funding costs of 8% invest in a new, low-risk project with a return of 5%? I think it arises because of the mixture of debt and equity (it would not be a problem in a pure-equity company) and it is exacerbated in the case of banks because of their high debt-equity ratio.

The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad