The “Geithner Put”, part 1

The details of the “Geithner Put” have been released.  It has two parts:  One to deal specifically with bad loans, the other to deal with other legacy assets (securitized yadda yadda).  In this post I will discuss the first part, dubbed the “Legacy Loans Program”.

The Treasury helpfully provides an example, which I reproduce here:

Step 1: If a bank has a pool of residential mortgages with $100 face value that it is seeking to divest, the bank would approach the FDIC.

Step 2: The FDIC would determine, according to the above process, that they would be willing to leverage the pool at a 6-to-1 debt-to-equity ratio.

Step 3: The pool would then be auctioned by the FDIC, with several private sector bidders submitting bids. The highest bid from the private sector – in this example, $84 – would be the winner and would form a Public-Private Investment Fund to purchase the pool of mortgages.

Step 4: Of this $84 purchase price, the FDIC would provide guarantees for $72 of financing, leaving $12 of equity.

Step 5: The Treasury would then provide 50% of the equity funding required on a side-by-side basis with the investor. In this example, Treasury would invest approximately $6, with the private investor contributing $6.

Step 6: The private investor would then manage the servicing of the asset pool and the timing of its disposition on an ongoing basis – using asset managers approved and subject to oversight by the FDIC.

Let’s flesh this out by repeating it 100 times.  So say a bank has 100 of these $100 loan pools.  And just by way of example, suppose half of them are actually worth $100 and half of them are actually worth zero, and nobody knows which are which.  (These numbers are made up but the principle is sound.  Nobody knows what the assets are really worth because it depends on future events, like who defaults on their mortgages.)

Thus, on average the pools are worth $50 each and the true value of all 100 pools is $5000.

The FDIC provides 6:1 leverage to purchase each pool, and some investor (e.g., a private equity firm) takes them up on it, bidding $84 apiece.  Between the FDIC leverage and the Treasury matching funds, the private equity firm thus offers $8400 for all 100 pools but only puts in $600 of its own money.

Half of the pools wind up worthless, so the investor loses $300 total on those.  But the other half wind up worth $100 each for a $16 profit.  $16 times 50 pools equals $800 total profit which is split 1:1 with the Treasury.  So the investor gains $400 on these winning pools.  A $400 gain plus a $300 loss equals a $100 net gain, so the investor risked $600 to make $100, a tidy 16.7% return.

The bank unloaded assets worth $5000 for $8400.  So the private investor gained $100, the Treasury gained $100, and the bank gained $3400.  Somebody must therefore have lost $3600…

…and that would be the FDIC, who was so foolish as to offer 6:1 leverage to purchase assets with a 50% chance of being worthless.   But no worries.  As long as the FDIC has more expertise in evaluating the risk of toxic assets than the entire private equity and banking worlds combined, there is no way they could be taken to the cleaners like this.  What could possibly go wrong?

Update 19:30

Wow, my site got Felixdotted (SlashSalmoned?).  In response to some of the comments there and elsewhere:

Yes, with the auction process, the assets will get bid up to the point where private equity (and Treasury alongside them) do not make such massive profits.  But in the process, the outcome becomes even sweeter for the bank and worse for the FDIC.  This appears to be the entire point of the exercise.

And yes, the FDIC is funded by the banking industry itself.  Or has been so far.

Since the day-to-day mission of the folks at FDIC is concered with defending the integrity of their insurance fund — that is why they seize banks in the first place — this proposal is likely anathema to the culture of the organization.  And I did read somewhere about rumors that some people at FDIC were objecting to this plan.  (Sorry, I lost the reference.)  This rings true to me.

I have more to say in Part 2.

Another update, next morning

Yes, yes, my model where half the pools are worth zero and half are worth $100 is totally unrealistic.  It was not meant to be realistic; it was meant to be illustrative.  I am of the opinion that most people, even intelligent laypeople, do not know that non-recourse high-leverage loans are equivalent to a put option.  I suspect many of them do not even know or care what a put option is.

Yes, a binomial distribution is oversimplified and also worst case, since it maximizes the variance and therefore the value of the “embedded put”.  But it also yields to a two-paragraph analysis that requires no ability to understand what my previous sentence even means.  My intention was to explain the principle in simple terms and to do a little exploration of who exactly will be left holding the bag.

…but if you want to see a more elaborate model, I gave it a whirl. And then I blathered on some more in another post.

48 comments to The “Geithner Put”, part 1

  • DCRogers

    You forgot Step 0: Straw bidder has line of credit with offering bank to fund initial “private” equity position, with understanding that bidding will be forced over bank’s break-even point.

  • grr

    if you are going to use numbers, please use numbers that are reasonable.

    the numbers you have chosen (50% chance of being worth zero and 50% chance of being worth 100) are cooked up to maximize public losses and maximize private gains. in reality, no pool of loans is going to be worth zero and no pool is going to be worth 100, and in reality the bids are going to track the final values at least a little bit.

    there’s no doubt that this is intended as a subsidy and the most likely case is that it is a subsidy.

    however, what we are paying for here (if we are paying for anything) is some clarity about marks. if the marks are really that bad (pools of loans marked at 50 but they could be 0 or 100) then we are going to pay for that clarity one way or another.

  • grr –

    I started with the Treasury’s numbers and made up probabilities to illustrate the principle. But in reality, the probability distributions would be far more complex, and the price paid would be an output of the calculation, not an input. My point is very simple: Regardless of the probabilities you assign, non-recourse high-leverage loans allow investors to overpay for the assets and still realize a profit. They encourage (i.e. reward) private investors for overpaying for the assets, and this remains true no matter what “realistic” numbers you assign.

    Consequently, the idea that this will provide “clarity about marks” is darkly humorous. The marks created by this exercise will be inflated by construction. This is possibly the intent.

    But what this part of the plan does most definitively is to transfer risk from the most troubled banks onto the FDIC. Why Treasury wants to do that is a question somebody ought to ask Timothy Geithner.

  • ceptri

    But isn’t the FDIC funded by an effective “tax” on the banks. So wouldn’t they just raise the bank tax (or wait since the FDIC historically builds up a reserve over time) until the $3600 is repayed? Sounds like a simple way to effectively have the banks slowly dig themselves out of this problem. That’s assuming that the government lets the FDIC go into debt for the next several years. The only real downside I see is that it is slightly unfair to banks that are in good shape, since they will have to pay the increase FDIC fee even thought they didn’t participate. But a lot of good actors, namely credit unions, aren’t part of the FDIC, it doesn’t seem that unfair to me.

  • ceptri –

    Yes, the more I think about this plan (or this part of it, since part 2 is a little different), the more I think the intention is to transfer the risk for these assets to the FDIC. And to some extent, the FDIC already bears this risk…

    This structure is essentially begging private investors to take advantage of the FDIC for the benefit of the bad banks. Assuming the FDIC is actually required to repay the Treasury via the fees it collects, it does only harm all the sound banks to help the troubled behemoths. But I guess that is better than sticking it to the taxpayer.

    How long would it take the FDIC to repay hundreds of billions by collecting fees?

  • grr

    nemo -

    i responded at MR but will respond here as well, quicker.

    i understand that you were illustrating a point and i agree with the point. this is a subsidy. i thought that fact was clear to everyone, and so i didn’t think anyone would bother blogging about it, so i mistakenly thought that your point was that the size of the subsidy was X.

    now i am not against subsidies. they are baked into ‘too big to fail’, if we guarantee the bank’s debt. i do want to know what i am getting for the money.

    now you disagree with my comment that this will bring clarity about marks. clearly the marks will be inflated by construction. well, then at least this gives us an upper bound for the mark. it also gives us a decent way of estimating a lower bound for the mark if you are willing to use black scholes to estimate the value of the geithner put or some other means to bound it.

    at this point there are questions about whether and what to nationalize — and in the middle of this there is a big problem with valuation of illiquid securities. nationalization is going to be costly to the taxpayer or the debtor and it is going to ruin these institutions. so i would like to do it only where it is truly justified, and i believe that the market for these securities is in fact broken and not a good guide to value. i am not even sure that market values for these securities are too low at the moment — i am just saying that they are not a clear signal. so i don’t mind paying something for a signal.

  • hemantv

    Let’s try something a bit less pessimistic – say half at $90(Good) and half at $30(Bad), The way I figure:
    On Good both Investor and Treasury make $300
    On Bad both lose $300 for Net zero
    Bank makes $2400, hence FDIC loses $2400 i.e. we the taxpayers lose $2100.
    The point that FDIC is going to lose in most cases is valid + it doesn’t seem to have any upside at all
    However, my example does suggest that the investors are likely to bid well below $84 to give themselves better upside potential.

  • The FDIC is implicitly insured by the federal government, so any losses beyond the FDIC’s ability to pay will be borne by the taxpayer. The savings and loan bailout of the 1980s was basically a $200 billion infusion of cash into FSLIC, the FDIC’s former savings and loan counterpart, to cover its losses. As grr notes, a level of subsidy was always likely to be the case, but the question is whether the current program increases or decreases the cost to the taxpayer as opposed to simple temporary nationalization.

  • mattino

    Very interesting and helpful but one question — the 16.7% return you point to for private equity in your example could be a return realized only over a period of years, right? Depedning on the time frame that means it may hardly be some ample annualized return at all, either for PE or for Treasury which tracks the PE return. Which seems to point once more to the shift of costs to FDIC as the main thrust here. How does this time frame point affect your analysis? Thanks

  • Wagster

    Nemo, your critique depends on the private market being totally clueless on how to price these assets. I don’t think the laws of credit have been suspended — we can make projections on who is likely to default without the 0% certainty you are taking as an assumption.

    As the range of uncertainty narrows, the premium that rational investors would place over the real price narrows too. In fact, if the range of downside uncertainty is less than the percentage of the total investment which the private investor has at risk (17%) then the rational investor would not put any premium over what he perceived to be the real price at all.

    Yes, in real life it is likely we will overpay, but not by the amount you imply… it’ll be maybe a percentage point or two. Before you scoff, model it out with a realistic distribution of outcomes. You might be surprised.

  • veritatis cupitor

    So, could a bank with a toxic pool of residential mortgages put them up for auction, bid on that very pool, and actually come out ahead?

  • marksf

    “Half of the pools wind up worthless, so the investor loses $300 total on those. But the other half wind up worth $100 each for a $16 profit. $16 times 50 pools equals $800 total profit which is split 1:1 with the Treasury. So the investor gains $400 on these winning pools. A $400 gain plus a $300 loss equals a $100 net gain, so the investor risked $600 to make $100, a tidy 16.7% return.”

    Is that the way it works? After reading the summary, I would have though this:

    The fund recoups $5000, for a $3400 loss. $5000 is not enough to cover the $7200 FDIC insured financing. Treasury and investors get $0. Bank gets the $5000, and the FDIC has to some up with $2200 from “somewhere” to make the bank whole.

    In this case it’s dramatically a less appealing “put”. I sure hope it’s that way.

    Depends on if the “put” in on a per pool basis, or for the entire fund.

  • DTM

    You appear to have omitted the Guarantee Fees that the FDIC will charge in exchange for guaranteeing the loans, which will be senior debt secured by the assets in the pool. So, some of that $400 from the winning pools will come back to the FDIC to help pay for its losses on its guarantees in the losing pools. It is also worth noting that the allowed leverage may be less than 6-1 for a particular pool, based on the FDIC’s assessment of the risks of that pool (with help from the Third Party Valuation Firm).

    Of course you have structured this hypothetical in a way that makes it impossible for the FDIC not to take a large loss on its guarantees regardless of what fees it tried to charge. But in an example where the average price paid for the assets was closer to their average real value, and where the leveraging was lower for riskier pools, it is possible that the Guarantee Fees would in fact cover the FDIC’s losses on losing pools.

  • foss

    wow, been awhile since I posted, but did you see that via Felix you got linked on Krugman’s blog?

    I mean, I knew you were a smart guy and a good writer, but a Nobel laureate just gave you props.
    I don’t see the world the way Krugman does, but he is a smart guy, and his praise is nothing to sneeze at.

    Congrats – you deserve it and keep up the god work.

    And for the record, Part 2 is actually even better, because no one is paying any attention to it!

  • dumbmoney

    Instead of this all or nothing assumption, a normal distribution would be more realistic. Keeping your numbers, assume there is an equal chance of the asset being worth $0, $50, or $100. Still an average value of $50. If the investor Bid $84 (of which $6 is his) he would lose $6 in 2/3rds of the case and make $8 in 1/3rd of the case or would end up with a loss of $133 on a $600 investment (or a 22% loss), which isn’t as rosy as your scenario.

  • foss

    Wagster – I doubt it would be as little as a percentage or two, but even so – entire fund investment companies/funds/strategies have been built on ‘just a percentage or two more’

    Marksf – your error is that you are looking at all of the loan pools in aggregate – that isn’t how the financing works. The non recourse funding is provided for each separate loan pool. So each time a loan pool comes up worthless, the private investor loses everything on that pool, as does the Treasury and FDIC. Each time a loan pool comes up profitable, the FDIC gets its money back (plus a small amount of interest) while the Treasury and Private Investor profit. So each time a pool blows up, the FDIC gets cleaned out at potentially a loss rate of 6x that of the private investor. Each time it comes up roses, the Treasury and Private Investor come up with gains, at 6:1 leverage, while the FDIC gets its money back (plus a tiny bit of interest).

    Now, if you could get 0 leverage on the downside, but 6:1 leverage on the upside, would you take it?

    If your answer is yes, you would love to have a time back to 2003. I could have found you a nice 100,000 house, and you could get infinite leverage with no money down. The value of the house goes up, you get huge upside due to the leverage, your property goes down in value, you walk away -you lose some pride, some of your credit score, and not much else.

    If this story looks familiar its because our solution to this toxic waste problem is to play the same game we played from 2003-2008, but this time, for higher stakes, and every single taxpayer has to play…

  • jschwarz

    Looking at probabilities less than .5 is interesting. If the probability of the pool being worth $100 is .3 then according to my calculations the break even point for the investor is $86. That is, if they buy pools for $86 they will break even (as will the Treasury on its share of the equity) but the FDIC will suffer on average about an $52 loss. I picked .3 because I recall reading somewhere that there were some assets being carried on the books at 90 cents (on the dollar) that could be sold for 30 cents (on the dollar). This is an enormous subsidy to the banks.

    It’s hard for me to imagine that Geitner and Sumner haven’t done similar computations himself. So why did I just see Sumner on Lehrer saying that the plan is designed to determine accurate prices? My cynical belief, without any evidence, is that they are making a political calculation about how to put more money into the banks.

  • Wow this is a lot of comments… I may have to do a new post to respond to some of them.

    But briefly:

    dumbmoney –

    You make a fair point. But as I mentioned in an earlier comment, the $84 is just part of this example. In real life, that price would itself be determined by these probabilities; or more precisely, by the estimates of these probabilities by the private equity firms. In your example, where each asset has equal odds of being worth $0, $50, or $100, no investor would bid $84. But one might bid $70. Then on 2/3 of the assets he loses $6 and on 1/3 of the assets he earns $15 ($30 split with the Treasury). So the total rake is again $100, and the investor still overpayed by 40%.

    You and Wagster are right that the actual probability distribution makes a big difference, and I admit I have not run the numbers for a “realistic” scenario because I do not know what is realistic. My intuition is that the uncertainty (i.e. variance) is quite high right now, because the true value of these things depends on the trajectory of the broader economy, and that is more uncertain today than at any point in our lifetimes. The expectation value also figures into it, and I suspect that is pretty low for a lot of these cases. But again I do not have “realistic” hard numbers.

    But I think these are the right questions to be asking, and I do not believe that enough people are asking them (yet).

  • dumbmoney

    By the way I do agree that there are a lot of troubling questions with this plan. I just think you are focusing in the wrong place. To me the biggest problem that the transactions probably will not be done at arms length. Which means that the bank selling the assets has some relation to the party bidding on them. And in this case the bidder pays an artificially high amount essentially wiping the assets off the books of the bank and putting it on the FDIC/Treasury/tax payer. Considering that one of the principle players that is said to be on board with this is Black Rock (which surprise surprise is half owned by Bank Of America). So in your example black rock bids $100 on every pool, and looses $6 on half of them. The bank makes an average of $50 or double what the assets are worth, Black Rock takes the average loss of $3. So essentially the bank just unloaded their pools for $97 each, and made a whopping profit while transferring the loss to the government. That’s the true danger here, and I haven’t seen anyone talk about how its going to be prevented.

  • mattino –

    Very interesting and helpful but one question — the 16.7% return you point to for private equity in your example could be a return realized only over a period of years, right?

    Yes, and this is a good point. So DTM notes that I am ignoring the Guarantee Fees, and you note that I am ignoring the time value of money, and both of you are right… But I was really just making up an example to illustrate the effects of non-recourse high-leverage loans. (I also omitted the interest payments on the loans from the FDIC.)

    In reality, these assets are pools of mortgages, so they generate cash flow over time. Also, within a year or two, or whenever employment bottoms out, it should become a lot more clear how much each is really worth. At that point it should be possible to sell them for close to their “true” value in the marketplace. All of this would complicate the analysis.

    But I do not think it fundamentally changes the picture. Just assume that all of my figures are “net present values” and therefore they already take into account the time value of money. So are you willing to pay $600 today for something whose net present value is $700? (That is, it generates a 16.7% profit plus interest?) Sure, why not. Although I concede at this point we are nearing the limits of my layman’s understanding.

  • seancarmody

    Although, as Nemo notes, the numbers here are just a toy example, the whole plan will only “work” if the prices realised at auction are significantly higher than seen in the (admittedly limited and fairly dysfunctional) secondary market. At these “market” prices many banks would be insolvent, so Geithner and co. are attempting to engineer a solution that takes assets off bank balance sheets at a high enough price to keep the banks solvent. At lower prices, banks would not participant and would instead continue to mark to myth (in fact, even if this scheme generates higher prices, as it should, they still might be too low for some banks on the brink of insolvency).

    Regardless of the numbers used here, the structure effectively bundles the asset with a put option and must therefore be more valuable than the asset alone (the put must be worth something!). Now grr suggests that discovering the price of the asset + put bundle provides valuable price discovery. I disagree. The only way we could work from the bundled price to the value of the underlying asset would be to subtract the value of the put option, but that is far from easy! grr suggests applying Black-Scholes, but stretching idealised mathematical models to these assets is part of the reason we got into this mess.

    Ultimately whether this approach is a good idea or a bad idea comes down to how bad the underlying problems are with these assets. Geither and others view this as simply a liquidity problem. If only we could hold out for long enough, prices will normalise and everyone will be ok, even the FDIC as the Guarantee Fees will leave them ahead. On the other hand, if the problems are deeper and the assets ultimately do not recover, we hit marksf’s scenario where the FDIC is woefully undercapitalised to cover the losses and the good old taxpayer comes to the rescue. So which scenario is right? Who knows, but unless you are certain that the assets are going to be fine (which no one really should be), this is an awful leap of faith for the US taxpayer and, consequently, a get out of jail free card for banks who are in a position to take up the plan.

  • girloftheseas

    Sheesh! I don’t get this! Somebody give me the George Bush executive summary. You lose me at step 2: The FDIC would determine, according to the above process, that they would be willing to leverage the pool at a 6-to-1 debt-to-equity ratio. What’s “leverage the pool at a 6-to-1 debt-to-equity ratio” mean? They “would determine”? Sounds like it’s already pre-determined. I don’t really see where these numbers are coming from. I’m a lay person, for goodness sake, and I need more than a Swiss cheese explanation. I can do arithmetic, but you haven’t shown your work and spelled it out for idiots like me Instead of pulling numbers out of the air like rabbits out of a hat, connect the dots please!

  • amack


    New to your site, thank you for the open discussion. W/o deep knowledge of either economics or banking, but in the spirit of “Self-evidence”, I would like to raise two points for all in the discussion.

    1. Why has the view that the market is currently under-estimating the ‘real’ value of the assets in question become so broadly accepted?

    It would seem to me that at present, no one (with the possible exception of the banks themselves) is able to accurately judge the extent to which over-leveraged, negatively-motivated American mortgage holders with falling incomes are willing or able to repay their debt on an overbuilt home.

    The common assumption seems to be that the problem is not that Americans won’t pay, that instead it is today’s market that is sick or depressed. That with this room spinning around, it just can’t price things right. That some Geithner ‘market’s little helper’ will get it out of bed and back to pricing assets the way it used to – like back in ’07.

    Do we not agree that prices were highly inflated in 07, in 08? And that current marks are just as inflated? Before due to a housing bubble, now due to the US gov’t ‘wink wink’ that it will OVERPAY.

    2. If my assertion above is not true, i.e. if Americans will pay en masse and if these ‘assets’ are somehow temporarily ‘un-fairly’ priced, why can’t the banks raise the lion’s share of the financing necessary to get the 16.7% return discussed?

    Why is the American taxpayer the only investor in the world smart enough to assume the vast majority of time/money risk of revaluation of the ‘assets’. I must believe the American taxpayer is the only investor in the world ‘smart’ enough to load up on risk, but give away all the potential upside on very likely future NPLs.

    Why must OUR (taxpayer) collective balance be potentially destroyed while Citi (and crowd) share price soars?

    I continue to believe that we are confusing failure of one or more of the big banks with the failure of the banking system. And therefore, I am driven to support the view that the only real solution to this bubble is for the share price of the insolvent banks go to zero, the government take over ALL of the insolvent bank’s assets, fairly and ‘unfairly’ priced – good and bad, and we all take the appropriate medicine now.

    Please tell me why I’m nuts.

    Does not “too big to fail” = “too big to exist”?

  • Flytrapent

    Look, I know nothing about math and failed Econ in college — but this is what I don’t understand in your example –

    “And just by way of example, suppose half of them are actually worth $100 and half of them are actually worth zero”

    aren’t the assets you’re talking about houses? Even if a borrower defaults, isn’t the House still worth something? How can it be worth nothing?

    Please explain

  • Wagster

    I will add one other thing, Nemo. When we try to figure out how much a private investor would bid in these theoretical scenarios we tend to assume that the investor will bid up to the threshold of break-even. Of course, in real life it doesn’t happen that way. An investor is not going to take big risks for just an even chance of breaking even. The wider the spread of uncertainty, the more he will factor in for profit. Bill Gross of PIMCO says he is expecting returns in the teens. For the kind of risk he’s taking, that’s what he needs. So he’s going to factor that into his bidding. Also, he has some heavy expenses. The loan tapes for every security will have to be pored over. That’s at least a percentage or two. The government has neither expenses nor expectations of profit.

    After having thought about it overnight, I think these two factors will negate the premium private investors are likely to offer because of the capped losses. I don’t think he government would overpay at all, which is not to say they won’t lose money, and it’s not to say that the Geithner plan will fix things magically. It probably won’t. My guess is the banks will participate lightly on the selling side, wanting to continue in their zombie state. The real showdown is still to come.

  • jim_b

    Somebody help me out here. If I invest $600 and get back $400, how do I make $100?

  • Greenprof

    Nice post, good discussion, but I think we are missing the big picture.
    The underlying problem is that the banks have these assets on their books and they are not worth what the banks originally thought they would be. In your example, the true value of the assets is $5000. To continue the example, suppose the banks initially valued them at $10,000, assuming that house prices would always go up. The banks then have a loss of $5000 if they sell the assets at the true value, and this may make them insolvent. If they are allowed to, they may now go bankrupt. The government’s goal in this whole process is to avoid the bank going under. This will involve some kind of subsidy to cover the losses. In this example, the FDIC ends up paying $3600 and the bank accepts losses of $1400. If that is enough to get the banks solvent again, then the plan can be considered a success. To compare it with other options such as nationalization, we need to look at the end cost to the government – in this case the FDIC takes a large share, but not all, of the bank’s losses. If they can cover those with higher bank fees in the future it sounds like a good deal to me.

  • girloftheseas –

    What’s “leverage the pool at a 6-to-1 debt-to-equity ratio” mean?

    It means the fund puts in 1/7 of the money and then the FDIC provides loans for the remaining 6/7. So if the fund bids $84, 1/7 of that ($12) is the fund’s own money and 6/7 ($72) is borrowed from the FDIC. The $12 piece is called the “equity portion”, and it is first in line to bear any losses on the purchase. (Put another way, when the asset is ultimately sold, the fund has to repay the $72 loan before it can get back its own $12 or earn any profit.) The other $72 is called the “debt portion”, and their ratio is called the “debt-to-equity ratio”.

    Put yet another way, when you buy a house with 20% down, your debt-to-equity ratio is 4:1.

    In addition, the Treasury jumps in to take 1/2 of the equity position. So 1/2 of the $12 comes from the private investor, and the other half comes from (and belongs to) the taxpayer.

    I encourage you to work through the example and ask more questions if necessary. The whole point of this post is to let non-experts get a real understanding of how non-recourse loans can be a form of subsidy.

    jim_b –

    Somebody help me out here. If I invest $600 and get back $400, how do I make $100?

    My phrasing may not have been clear; I do tend to bounce back and forth between “gross” and “net” numbers.

    But the numbers are correct. You invested $600. Half of it ($300) was lost. On the other half, you got back your $300 and also made a $400 profit (that’s the “leverage” at work). So in your terms, you invested $600 and got back $700 for a net profit of $100.

  • dm

    I have a question please: Why is it that we have so little idea of the value of these assets? I know the specific value of a specific asset may be hard to determine (because mortgages were bundled up, sliced and sold) but some things are roughly knowable and should help us understand the value of these assets on average at least: The number of people that are or are likely to default on their mortgage is not a random number–I think 90%+ or so of mortgages are not in default,, the numbers of the number of homes with mortgages pre-2000 (presumably most of these are not upside down), etc. Shouldn’t it be fairly straightforward to project some reasonable ranges of the average value for these assets so we are not at such risk as the “could be $0, could be $100″ scenario?

  • blue_diamond

    can someone please explain non-recourse loan? i don’t understand how they can loot the FDIC w/ profit.

    A lender lends money to the Fund and FDIC guarantees the loan to the Fund. The Fund loses money on half of the loans and makes money on other half. Let’s say, when all is said and done, the net on the Fund is negative: there are more bad loans than good loans for that Fund’s pool.

    In the post above, it is implying that the Fund will split up the profit from the good loans with the Treasury and does not intend to pay the lender at all???

    I read the 5 page white paper from the Treas. I am guessing i have this all wrong.


  • maxwellthedog

    Nice post, but quite misleading for two reasons. (For that matter, Krugman’s is equally misleading, but he tries to be clever about it as I will show).

    First, while it is true that there is a subsidy from the government (via a put that comes from a non-recourse loan), that value of that put is not nearly as big as you say. And there is good reason to expect it to diminish to almost nothing over a very short period of time.

    To prove my point, consider a counter-example to your example. Imagine there are 100 loans and they are each worth *exactly* 50 (variance of zero). What would private investors pay? For simplicity’s sake, let’s assume required return to the investors is zero. Then each investor would pay exactly… 50.

    Investors pay $50 (of which they put up about $3.50, matched by the Treasury’s $3.50, levered by $43 from the FDIC – 6 to 1). The bonds are worth $50, so investors get back their $3.50, the Treasury gets back its $3.50, and the FDIC gets back its $43. There is no subsidy, banks sell assets for what they are worth, and investors make their required return (in this case, 0%).

    Now imagine a case where the assets are worth either $25 or $75. Still an average value of $50. What do investors bid? The answer is about $58, for an implied subsidy of about $8. In your original example, if you assume the required return is 0%, then the price paid by investors would be about $86, for an implied subsidy of $36.

    The point you casually toss aside (that a binomial distribution is a worst-case scenario) is vitally important here—if you argue that there is a significant subsidy, then that subsidy is directly correlated to the level of uncertainty around prices. As price uncertainty falls, the subsidy drops dramatically. Case in point—just reducing uncertainty by 50% from your example to my second example reduces the subsidy from $36 to $8—more than 77%!

    (This is why Krugman is also misleading—he used the same 100 point distribution in his example, but shifts it from $0 – $100 to $50 – $150. Presumably the shift is to avoid the problem someone in this thread pointed out—that loans will almost never be worth zero. But they will never be worth $150 either! Bond prices top out around $100 (plus or minus). So Krugman’s example tried to make the same economic point as yours by using the same enormous variance, but he shoots himself in the foot. If he used $50 to $100 as his range, his implied subsidy would be dramatically smaller—and his argument dramatically weaker)

    Your example also misses a larger point. Imagine I am an investor facing the case you outline in your original example—loans are worth either $0 or $100. Everybody is bidding $84 for their 17% return. What do I do?

    The answer is: I spend all of my time trying to identify which loans are worth zero. Any edge I can gain at all in making this determination will dramatically improve my returns. For example, if I can identify just 10% of the loans that will be worth zero, and only bid on the remaining 90, them my return goes from 17% to 23%. That is a lot of incremental return for my work. In fact, what I can actually do is bid a *touch* higher than everyone else, win every auction I want, and still earn my higher return.

    What is the impact of my actions on the market? On 90% of the auctions where I have no edge, I bid market price plus a little, and win. And on auctions of bonds where I expect there to be a loss, I don’t bid, and the marginal price is a little lower. Repeat this process over hundreds of market participants, and you can see how prices will start to gravitate towards “correct” levels, even with highly imperfect information.

    The final point—this process happens on an auction-by-auction basis. You could have a market where bonds are worth $100 or $0 (about as likely as them all being worth exactly $50), but if market participants are bidding the right price for each pool, then the effective subsidy is still zero. Prices paid will be the correct prices (or close to them) and the government incurs a minimal loss.

    If you actually look at the real mortgage market, you would see that there is a range of prices—many of which are $30 to $50, depending on the vintage, seniority, and deal. The worst of the pools are from 2005-2007, so they are already aged by two to four years. Average mortgage lives are seven to ten years so a substantial portion of the ugliest pools are already paid down (or defaulted). On the remaining outstanding mortgages, the behavior is beginning to become more predictable (i.e. if they haven’t defaulted already, the chance they default in the future is a lot less). Lower prices, smaller remaining pools, longer aging, and greater certainty about loan performance means much better pricing accuracy. I would bet the actually government subsidy paid is only a couple of percent relative to loan prices. And that this subsidy is a lot less than what the government would fork out in a complete nationalization.

    There is the larger question of whether the banks will actually want to (or be able to) sell. Undoubtedly there are a percent that are not solvent if they mark to market. But with this scheme, the Fed and Treasury will begin to have a better set of prices that they can take to a bank and use for determining if that bank is worth saving or liquidating.

    This plan might not work. It all depends on participation. But it is a pretty good idea that can be tried at a relatively low cost to the government.

  • maxwellthedog

    Sorry– typo. It should say “if I can identify just 10% of the loans that will be worth zero, and bid on the remaining 95…”

  • cityhall

    “am of the opinion that most people, even intelligent laypeople, do not know that non-recourse high-leverage loans are equivalent to a put option. I suspect many of them do not even know or care what a put option is.”

    It’s equivalent to a call option, not a put. If the asset goes up, you make money. If it goes down, your losses are capped near zero.

    The “Geithner put” terminology means that banks themselves have been given an implicit (becoming more explicit) put option on their own assets, allowing them to sell the assets to the taxpayer through various intentionally opaque schemes at above-market prices.

    The Treasury is settling the puts it gave the banks by giving away calls to the hedgies, who will in turn pay above fair value for the bank assets knowing Uncle Sam will eat most of the losses. Whether its a reasonable use of taxpayer money will depend on how aggressively it gets arbitraged.

  • seancarmody

    @Flytrapent: The extremes of $0 and $100 are for illustrative purposes only. You are correct that it would be hard for a simple pool of mortgages to be worth zero because some borrowers would not default and even where defaults occur, the houses would be worth something (although, perhaps not very much. Moving to other sorts of assets, such as asset-backed CDOs (i.e. sliced and diced mortgage pools), it is quite possible for these to be worth zero.

    @Greenprof: the difference between the Geithner plan and nationalisation is who bears the losses. The Geithner plan effectively bails out a bank’s creditors (such as bondholders) and even retains a bit of value for equity holders, while shifting most of the risk to the FDIC. Under nationalisation, equity holders would lose everything and most creditors would absorb some losses too as part of writing down the bank’s assets. While the Government would assume risk from that point, most of the pain should be taken by at the outset.

    @maxwellthedog: you are certainly correct that reducing uncertainty reduces the value of the put and therefore the “asset + put” price should get closer to the value of the asset by itself. However, even with more realistic analysis and modelling of these assets, there is an enormous amount of uncertainty remaining as to how much further house prices can fall, how much further unemployment could increase and how much worse these assets can get. That means that the put is worth something, not zero. Currently, uncertainty in the market is still very high, which means that the put should be worth quite a lot! Is it really low cost to the Government? One answer turns on whether the value of the put is more or less than the Guarantee Fee. But, like any insurance policy, the answer will eventually show up in claims. In a sense it doesn’t really matter whether the price was right in the end: if there are few claims on the Guarantee in the end history will judge it a good plan, if the claims are huge it will be judged a disaster. Right now, we can only guess.

  • cityhall –

    Congratulations, you have discovered put/call parity.

    You can view this as an option to acquire the asset which you plan to exercise only if it is profitable; and yes, that is a call option.

    But you can also view it as actually purchasing the asset, while having the option to sell it back at a certain minimum price. That is a put option.

    Since the fund is, in fact, purchasing the asset, and the non-recourse loan allows them to “put” the asset back to the FDIC and thereby discharge the loan, I think viewing it as a put option is more natural. But both formulations are correct, because they are equivalent.

  • seancarmody

    @cityhall: as Nemo says, put-call parity essentially says “asset + put = call + cash”. The bank sells the asset and the FDIC provides the put, so the investor gets a bundle which, as you say, effectively means they have bought a call. It’s being referred to as “Geithner’s Put” because it’s his idea that the FDIC should provide a put option. The Guarantee Fee they charge is basically the option premium for that put.

  • maxwellthedog

    @seancarmody– i agree, the put has value. but my point was twofold: first, uncertainty matters. While there is certainly uncertainty in the market, it is not nearly as significant as posited by the post here or by Krugman. If someone is looking at buying a bond priced at 20 cents, the downside/upside asymmetry is significant and the range of outcomes (and therefore the value of the put) are a lot smaller than in the previous examples. and as you say, at the end of the day, the value of the put (or expected claims against the Guarantee) has to be compared to the cost of other options available to the government. By inflating the expected cost of this program, we make it look less attractive than it should be. This program is incremental, provides information, and can be shut down if the government decides it is not the right path (unlike a nationalization, which is irreversible).

    My other point is that while it is easy to paint a simplistic picture of how this will work (with people blithely bidding $84 and earning a 17% return), that is absolutely not what will happen. Every investor has an incentive to be as accurate as possible in their pricing of bids. This is a crucial point– because of these incentives, we are guaranteed that 1) pricing information will be as accurate as people can possibly ascertain, and 2) that the value of the put is as low as possible. In fact, this is a nice alignment of incentives between what investors want and what the government wants.

    There is one other element– if you believe that mortgage assets are underpriced (as i do) then this is a great deal for the government and taxpayer. Sure, there is uncertainty about the economy, delinquencies, etc., but a large part of the problem in these markets has been the delevering of buyer’s balance sheets. There is a lot of supply and limited demand. It seems like we have reached a price where a lot of these assets are significantly underpriced (at least, that is what i hear from people who are actually buying right now). If there is a chance that the market is artificially depressed, then this is a much smarter route than nationalization or liquidation. I admit, there is a big element of hope in this side of the argument, but if the government can experiment in a controlled fashion like they are with this program, why not take this step first?

  • I really need some help with the VERY FIRST STEP:

    “Step 1: If a bank has a pool of residential mortgages with $100 face value that it is seeking to divest, the bank would approach the FDIC.”

    Why would a bank want to divest a pool of mortgages. The whole loan kind that we recognize?

    If they are part of the held to maturity portfolio, they aren’t marked to market. They are booked at amortized cost less loan loss reserves. The bank loan loss reserve process tends to lag expected ultimate losses, since it uses historical default ratios as key inputs.

    Read Short em Jim Chanos’s WSJ article re M2M…. if you want to see it in print.

    Also note that the Treasury proposal excludes CDO^2 or any asset backed security which holds anything other than whole loans as the underlying securities and must have originally been rated AAA.

    These pools of “toxic” assets don’t exist on regular bank balance sheets. City and MER/BAC have already had $300 billion of em ring fenced.

    They are all over the place, but not so much bank balance sheets.

    The only stuff that might qualify would be stuff that already has a huge haircut, like WFC’s pick a pay. However, I believe WFC intends to use these write downs to cover their own losses. Note that I’m not sure WFC is right, but they intended to take enough on WB to cover their own losses, so they need to keep the marked down assets and write them up as they write down their own.

    The Hartford has billions in private CMBS that are giving them fits. Not a bank and I’m very sympathetic to Hartford’s issues, but I can’t imagine this is for them.

    I dunno…..maybe it is just more for C, BAC, and JPM. They are 25% of the US banking system.

  • I suppose I predict that this will just be sitting out there and probably won’t actually be used.

    It will take a month to pick the managers, they get 30 days to raise capital, etc. Then we are into June and I don’t really see much happening anytime soon.

    However, it is POSSIBLE that the economy will start showing some signs of life by 3Q and this will be sitting out there to sop up assets when they must be sold and there is no reasonable bid.

    As long as there is a bid, a lot of the problem is solved. That and a few confirming transactions. It could keep us out of a Fisheresque deleveraging/deflationary spiral.

    Maybe I’m way too optimistic.

  • amack

    Did you believe S&P when the ratings were at AAA, well how about now?

    NEW YORK (Standard & Poor’s) March 24, 2009–Standard & Poor’s Ratings
    Services today lowered its ratings to ‘D’ on 245 classes of mortgage
    pass-through certificates from 243 U.S. subprime residential mortgage-backed
    securities (RMBS) transactions from various issuers. We removed one of the
    lowered ratings from CreditWatch with negative implications. In addition, we
    placed 29 ratings from four of the affected transactions on CreditWatch with
    negative implications. The ratings on 15 additional classes from Structured
    Asset Securities Corporation Mortgage Loan Trust 2007-BNC1 remain on
    CreditWatch negative. (see list).
    The downgrades reflect our assessment of principal write-downs on the
    affected classes during recent remittance periods. We lowered approximately
    97.96% of the ratings on the 245 defaulted classes from the ‘CCC’ or ‘CC’
    rating categories, and we lowered 100% of the ratings from a speculative-grade
    We expect to resolve the CreditWatch placements affecting these
    transactions after we complete our reviews of the underlying credit
    enhancement. Standard & Poor’s will continue to monitor its ratings on
    securities that experience principal write-downs and adjust the ratings as we
    deem appropriate.

    76 page report detailing the individual issues, mostly mezzanine tranches previously rated CC or CCC.

  • What Nemo describes is the subsidy hidden in the Geithner plan announced on March 23, 2009. My paper, “The Put Problem with Buying Toxic Assets” at argues that insolvent banks won’t sell their toxic assets without a big subsidy.

  • Fat Tony

    We’re mixing up risk-neutral and real worlds here. A security with a payoff of $100 or $0 with 50/50 probability is worth $50 only in a risk-neutral world; the real world price is far lower.

    With mixed success, lenders have been providing non-recourse debt to financial buyers for buyout deals for decades. How do we know if there’s a subsidy until we know the full terms, including rate, of the FDIC loan.

  • Fat Tony –

    I do not believe I am mixing up anything… I am providing an example where private equity could actually overpay for an asset and still make money. Yes, in the real world, the private equity firm would likely offer less than the expected value of the asset. But if the variance is large and the leverage is high, private equity will be willing to pay far more with these loans than without them, because the FDIC is assuming the tail risk. That is not free; in fact, it is likely to be extremely expensive. The private equity folks and banking folks know this, and I believe that is why they are so supportive of this plan.

    True, non-recourse loans are fine if the tail risk is not too large and the leverage is not too high. And true, we do not know whether that is the case here, yet. But if it were not, I doubt Bill Gross would be getting so excited. Perhaps I am too cynical.

    Regardless, this example shows that it is possible for the Geithner plan to become a truly massive subsidy, which I do not think is obvious to most people. (And the truly cynical would say that making it non-obvious is the point.) The detailed numbers matter a great deal.

  • grr

    Another issue that comes up is that the auction participants are not unconflicted. Some of them own the debt of banks and so it’s in their best interest that the banks have money to pay them back. Those participants have an incentive to overpay if the FDIC is providing a floor on their losses.

  • grr

    There’s one thing I wanted to write in my comment above but forgot.

    Remember a short time ago during the most recent TARP round with Citi, the treasury decided to convert some of its preferreds to common stock under the condition that other preferreds did so as well — essentially this was a synchronized action between Citibank and the Treasury to do an in-place debt-to-equity recapitalization which neither caused a credit event (avoiding credit derivative implications) nor put the Treasury in ownership of over 50% of Citibank stock (avoiding unintentional nationalization).

    AngryBear got what was going on:

    If this is right, then the Treasury needs explicit cooperation of bondholders going forward, or needs to move forward with nationalization. Not saying this is a payoff or anything, but again, the likely participants are far from unconflicted.

  • NYCycles

    The FDIC would only lose money if the debt issued by the purchaser of such legacy pools defaulted. If half the pool was worthless, then that wouldn’t be a $3,600 loss for the FDIC (at least not directly), but a loss for the initial purchaser. The only potential downside for the FDIC would therefore be a consequence of the holder of debt issued by initial purchaser of legacy pool being unable to pay off his or her debt, which in this economy is increasingly likely.

  • JAK

    I thought I was getting this, until the post above using the $25 and $75 pool values. By my (probably incorrect – I don’t know anything about econ!) calculations, the investor’s break-even point is at a $65.625 bid for each pool. I’m assuming the lender (=FDIC) has first dibs on cash inflows from each pool, but gets none of the profit, if any. Here’s my calculation for cash flow from a single pool that turns out to be a profitable investment (excuse me for including all this, but I want to be clear):
    V = total cash inflow from the profitable investment (=”true value” of the pool, $75 in this case)
    F = cash inflow to lender (FDIC), equal to the amount loaned to finance the deal (because none of the profit goes to the lender)
    R = fraction of profit going to investor (50 % or 0.5, because of equal equity stake between Treasury and investor)
    Then cash inflow to investor from purchase of a fortuitously profitable pool is I = (V-F)R.
    Cash outflow from the investor = O
    Net cash flow to investor (“profit”) = P = inflow – outflow = I – O = (V-F)R – O.

    Now, only a fraction (call it x) of the pools will actually turn out to be profitable. The unprofitable ones don’t yield any cash inflow, but investor still has to pay O for them. On average, cash inflow is then x(V-F)R, cash outflow is still O. So the net average cash flow P to investor is
    P = x(V-F)R – O

    After a few algebraic substitutions, using B for the bid price and L for the leverage ratio (=6 in the example above) I arrive at the expected return rate (= profit/outlay = P/O)
    P/O = x{V(1+L)/B – L} – 1
    (sorry, it’s a little difficult to read). For break-even, P/O = 0. Solving the above for the bid price B, I get B = V(1+L)/(1/x + L).

    Finally, doing the substitutions V = $75, L = 6, x = 0.5 (half of the pools will turn out to be profitable), I get B = $75 * (7/8) = $65.625, the bid price for the investor to break even. Of this, the investor and Treasury each put up about $4.69, and FDIC puts up $56.25. For pools where half are worth $100, and half worth $0, I find bid price = $87.5

    What am I doing wrong? I would greatly appreciate any help!

  • seancarmody

    @NYCycles the whole point is that the loans are non-recourse. So, if the value of the pools falls to the point where the investor’s (and Treasury’s) equity is eroded, the FDIC will have to pay up.

    Thinking about all of this again, I think that the Guarantee Fee is worth commenting on a bit more clearly. It is mentioned in a number of the comments, but not in the original analysis.

    Put simply, the analysis argues that auction prices will be inflated because of the non-recourse loans as the prices investors are paying are really the prices of an “asset + put option” bundle. This neglects the fact that the funding cost will build in a guarantee fee, so essentially investors are paying for the put option and they would subtract the cost of the guarantee from the price they are prepared to bid. So, they’d be actually be prepared to bid up to the value of “asset + put option – guarantee fee”. If the guaratee was “correctly” priced, this means that we would get true price discovery of the underlying asset. To the extent that the prices achieved are higher than could be achieved without the non-recourse loans this either means a) current market prices are excessively undervaluing assets (Geither/liquidity crisis view) or b) the Guarantee Fee significantly underprices the risk of the put option and the scheme represents a significant, albeit contingent, subsidy for banks selling assets.

Leave a Reply