I want to explain the Geithner Put for those without a finance background who want to understand how FDIC is going to get looted. (Warning: Will contain graphs, and be long and boring even by normal standards here.) Recap: Banks have “toxic waste” assets, which they want to sell. Geithner has created a public/private partnership, where FDIC will provide leverage, and government officials and hedge funds will provide some cash and set a bid on this waste. Self-evident has a useful follow-up to his “Geithner Put” post, which I blogged about yesterday. Nemo argues that the FDIC leverage provides a “put option” on the asset. He assumes a uniform distribution of values, and plays around with results. His results confirmed what he did on the back of the envelope – FDIC will be holding the bag.
Put option? What the hell is that? From Taibbi’s argument that we need more financial engineering literacy among the angry populists, I want to graph out what is happening – since I am a financial engineer and an angry populist. The Geithner plan will allow hedge funds to buy large chunks of bank toxic waste, and we’ll use denominations of $100 for each unit. How much to bid?
The problem for markets is we don’t know what these “toxic waste” assets are worth. If I was a hedge fund whiz-kid, or even a normal ho-hum banker worried about credit defaults on my portfolio of regular mortgages, I’d be assuming the value would look like a lognormal distribution (to be fair, at the end I’ll cover this, I’d normally assume that the losses, not values, look like a lognormal distribution – but these are not normal times). Just guessing, the “toxic waste” assets have a mu of 3.3 and a sigma of .5 – made-up, but reasonable. Hedge funds will use historical data to fit a distribution, and then play around with macroeconomic projects and other correlation matrices to fit it to the instruments at hand…it’ll look like what we have, except shifted a bit perhaps. But let’s go with these made-up numbers. Let’s draw one so you can visualize it:
As you can see, as with all lognormal graphs, the mean (red line) is larger than the median (green line). This will become important. So for every original $100 in toxic waste loans, it’ll have a distribution like above, where it is half likely than it will be less than $27. Some loans will be worth more than $100, so they get capped at $100 – this represents pre-payment risk, the idea that someone holding your liar’s loan won the lottery and paid off his mortgage in full.
So we believe that the toxic waste will have an average value, or its mean, of $30.65 – we should bid less than that. In a perfect market, we’d have to bid that. However let’s be crazy and bid $35 for this toxic waste. What would happen if we were a normal investor, who couldn’t leverage off FDIC? I’d have to actually pay $35 if I bid $35. Insane, right? Our payoff would look like this:
(Nerd disclaimer: The distribution is no longer in any kind of units on the y-axis, but is still to scale on the x axis.)
So that blue line is our profit. So if the asset is worth $35, our profit should be $0. It’s exactly what we paid. Note that blue line is at $0 at x = $35. If asset is worth $0, which it could be, we would lose our entire $35. If it was worth $100, we’d make $65. Make sure you follow this before moving on.
Now with the Geithner Plan, the FDIC provides 6-to-1 leverage. Which means we place a bid, and then only have to provide 1/7th of that bid. The FDIC leverage is non-recourse. So if we lose that 1/7th of the bid, we don’t lose any more money. FDIC takes care of the rest. If it goes up, we get the profit. Leverage!
Now what did I just say? Since we are bidding $35, we only need to place $5. This $5 is our “skin in the game”, as people are saying. FDIC covers the other $30. If it is worth $32, we lose $3. If it is worth $30, we lose all $5. If it is worth $10….we just lose our $5. FDIC is on the hook for the other $20. Look at the profit lines on this chart, which is the plan going forward:
Again: If the value of the asset is $30, we only lose our $5. What if the value is $29? We only lose $5 – however someone has to lose that extra $1 – and that is the FDIC. If the asset is $28? We only lose our $5 – that additional extra dollar is covered by the FDIC. If it is worth zero, someone still has to lose $35, and they do – we lose $5, the FDIC loses $30.
The Greenspan put was always a metaphor – the Geithner Put is an actual put option. If it goes up, we get it. Fantastic! If it goes down, our losses are capped. This is what having a put option looks like – under a certain point, we gain a dollar for every dollar we lose, so it nets equal. No value is created with options, so if we gain a dollar there then someone else must be losing a dollar, and that is FDIC. (note: I just fixed that graph so it is correct.)
Aside: If you think about it above, you may ask why would someone every sell a put option like that above? It is crazy – it always loses money and can never make money. Normally you sell it for some value – say $3 above. So if the asset is $30 or above, you make $3 ($30 is the “strike price” here). If it is $27, you break even. So giving away an option is the same as giving away money, in this case. Lots of fancy formulas have been written about how to calculate the value of an option – feel free to read the wikipedia linked above for a quick primer.
So what happens if we bid $35? Running that chart through some monte carlo math programming we expect to make $1.28. When I say expect, I mean on average. FDIC is expected to lose -$5.64.
Let’s look at a bunch of scenarios:
HF is for hedge fund. FDIC losses are in terms of total loan, not its leverage value – so it should actually be higher.
The $35 bid we just covered. But we are a mysterious and wise hedge fund, who has correctly figured out the average price – and with the government’s backing, we are “risk neutral”, and can bid the straight average. What if we bid the average? If we bid the average, we should expect no profit. If we agree to flip a coin a million times, and heads you get $1 and tails you get nothing, and you pay 50cents for each flip, over a million times you should end up with $0. Paying the average balances out. However, looking at the second line, since our downside is covered, we transfer out $3.66 for putting up $4.38, a nice 83% profit transferred straight out of the FDIC, which loses exactly that $3.66.
Now we believe markets are good though, right? Since we figured out that we can bid the average and make a profit, some other equally smart hedge fund is going to come along and simply bid a bit higher, making less profit but still some. We know this and do the same. So we both bid to the point where there aren’t any profits (line 3 of the chart). I assume this is what the administration has in mind when they say this will get the banks going again – since we are overbidding, we are giving the banks more money than we expect the asset to give us. And sure enough, we make no profits – and since that red line in the payout charts above has shifted to the right, there is more downside risk. And the FDIC transfers all that money straight to the banks.
Let’s go the other route. Maybe we get lucky! We underbid, the bank takes our bid, and the asset is worth a lot more than we expected. We make a massive profit, however FDIC is still expected to lose some money on some of the assets.
Now maybe you think this distribution is too low. Fine, it probably is. I’ve played around with the mean both higher, much higher, and much lower, and it is still a transfer from FDIC to the hedge funds. However, I’ve kept the variance relatively low; expanding that would increase the transfer, since there would be more tail risk.
So what’s the deal? Now I did a sneak – under normal circumstances, credit modeling usually assumes that the losses are lognormally distributed, not the portfolio value. What? That means that the value of the asset, according to normal credit modeling, should be flipped; the large bell should be around the high end, with a few stray values near zero. More importantly, the mean will be less than the median. This is what banks do when they look at their economic capital. A portfolio of highly rated bond instruments would look like this:
As you can see, almost all assets are above $70, with a small few here and there below, and a few more at $0 if you look really hard.
What happens if this is the case? I called that optimistic in the chart. The hedge fund bids the average, and it doesn’t make a profit (like it should) and the FDIC put option is worthless, since it rarely pays out. This is good for the taxpayers, banks, and the hedge fund. Which is to say, if it did look like this, the market wouldn’t need the government to jumpstart it – people would be lining up to get in on this deal. Holding this portfolio, modeled above, would be a godsend in the current environment. More generally, I’d be very, very surprised if this is what it looked like – because if it did, everyone would want to snatch it up.
I’ve played around with this in matlab for a little bit (I can post the code if anyone is interested); the real issue with determining this in this tail median/mean issue. If you think that all these subprime loans are essentially good, with a few stray bad ones in there, then this is a great deal for taxpayers. This would be surprising, since nobody wants to go near these things. If you think that these subprime loans are shit, and there will only be a few good ones in there, then we are essentially transferring money to banks and hedge funds straight out of FDIC.
Lame.
Oh – Last two things as a quick update:
1) When I say “hedge fund” above, I mean the public-private half-Treasury, half-Private (hedge funds, private equity, connected insiders, etc.) creation. So technically half the profits go to the Treasury. Unless the bids come in very low, which seems against the idea of what they are doing, half the hedge fund profits above won’t cover the expected losses of FDIC, so it’s a net loss to the Government.
2) And in case you are wondering, FDIC gets its money from taxing banks. Read about the “emergency” fee they are charging all banks in order to recapitalize for 2009, including this move, to supplement the line of credit from the government. Which is it say, we are being joined by medium and small banks, the thousands of them that were responsible with their lending and banking practices, in paying large taxes to bail out their larger, more reckless, and more well-connected competition. You know, those Too Big To Fail Banks, which are necessary because they give us better services at lower risk. Viva crony Financial Capitalism!
Another Update
Great comments. I didn’t want to dig too much into the microstructure of how these are set up in this post, because I wanted to pitch it at a level of people who are interested in learning about derivatives and put options without any background. A few things.
3) Many people are bringing up that the FDIC sets the leverage. Now if the leverage goes down, all that does is moves the light blue FDIC Put line to the left – it lowers the “strike price”; the dark blue hedge fund line moves appropriately. The option is worth less, as it is strictly increasing in strike price, but it is still a put option.
Now I have leverage exogenous to this above – the distribution sets the bid, and the bid and leverage set the payouts. Will leverage also depend on the distribution? Will the FDIC be able to handle political pressures from Geithner to make sure this works, and will it understand these mortgages better than the private equity and hedge fund folks? As self-evident said “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?” Yup.
4) Some are saying that the assets could/will be more than projected above. True. What didn’t occur to me until I started playing around with this model is how much the FDIC gets hosed even if the assets are great. (I made a lot of charts, but this is already too long). Let’s say the average value is 74.6 out of 100 – very optimistic (mu = 4.3, sigma = .2). The hedge fund knows the distribution in advance, because MBA programs are that good, and bids the average. No profit right? Except they still make $1.63, a nice 15% profit – and that is straight from FDIC.
Wait! Why? The assets turned out fine! Because, without getting too technical, the mean is larger than the median, and the volatility has value in an option. Look at the graph – they are bidding, in that red line, above where most of the values are. So they are chasing all the upside (mean), while the FDIC is left looking after downside, which happens to be where most of the distribution is (median). Also because the volatility here has value to them in the form of the option, regardless of expected value. I will post the matlab code that takes in the leverage, bid, and parameters and gives the estimates and draws the graph, if anyone is interested.
So if you think digging through all these mortgages involves just rooting out a few bad ones (median is greater than the mean) we are fine; if you think it involves trying to find a few good ones among all the crap (mean > median), then the FDIC gets taken to the cleaners.
5) Followups What also surprised me is that the put option, under some assumptions, is more valuable than the banking assets, and I argue the hedge funds may just try to maximize that instead here (somewhat technical, I tried to keep it easy).
For those who say that nationalization is also a put option, and the money comes from the FDIC one way or the other, I talk about how the path dependency does matter for our country.
Thanks for stopping by!
Rortybomb 
EXCELLENT, excellent post Mike! Maybe now I understand a bit about this “leveraging” business. But god do I love me some good lognormal distributions!
Basically what you are saying, distilled down to plain conversational english, is that the entire Treasury plan for rescuing the troubled markets is an elaborate scheme to transfer tax revenue from the US government into the hands of fund managers, because as long as the people’s 401k’s stop dropping then they will be placated. The money transfer seems somewhat more palatable when it goes through this long drawn-out process than if the cash were just handed outright, but in the end it’s just laundering. Which is a crime of course!
Hey thanks! Leveraging means a couple different things depending on when it is used – normally it just means having a larger position than you have put up the cash for.
Distributions are fun! It’ll probably have larger tails, and could take a different form (gamma?), but when the hedge funds bid, it’ll probably have something that looks like above, and will certainly have the same logic as how to bid once they have their distribution.
Yes. A lot of the defenders of the current Plan are more or less saying “We need to transfer $1T to the banking sector, so doing it this way involves less complications than nationalizing, which won’t save us much money.” I’m going to deal with that in a post later this week, but I do too think it matters.
Oh, I just wrote a short update – in addition to taxpayers, the government will also be paying for this by taxing small and medium sized banks. Like your local favorite small town one. Taxing from small firms to transfer to collapsed well-connected inside ones. Shameful. If they were being taxed for the nationalization of these firms, at least they wouldn’t have to compete against a steroid zombie bank, who no matter how you shotgun them, keep coming back….
Thanks for reading and commenting!
Pingback: self-evident » Fun with uniform distributions
I fixed the third graph, the one with the FDIC line, and added a clearer understanding of what a “put option” is – and why the Geithner Put is an actual put option here.
This is not a metaphor. This is real.
Here’s my naive i-don’t-understand-the-econ question: is this coming as a surprise to Geithner, Obama, FDIC, or anyone else? I mean, I think they’re going into this with their eyes open and the whole purpose (politically, if not economically) is to take it in the ass to the tune of a few hundred billion dollars in exchange for not having to take the explicit step of nationalizing large, failing banks.
It’s a bad deal for us, it’s a bad deal for the government, it’s a bad deal for everyone who isn’t Citi. But I think that’s the point. Unduly optimistic projections and estimates aside, do you think Treasury really thinks this is going to turn out well? I read it – again, relying on the econ interpretations of others – as just another round of sword-falling, a “what’s another $500 billion at this point?” maneuver. I mean, they stopped playing with real money so long ago. It’s all Monopoly money and abstract spending out of future budgets now.
Yes they know what is happening. If they don’t thats even more scary.
Lets also be honest that if the FDIC becomes undercapitialized taxpayers will need to refill its coffers (that’s what happened to the FSLIC).
The thing that worries me is that the administration is openly lying to us. They talk about getting market prices to keep from overpaying for the assets while they know very well that they are setting up a market that will guarantee overpayment.
Another issue is that by not nationalizing the banks we are allowing the banks to determine which assets to sell. Now some are afraid that the banks will dump the most toxic ones, I have the opposite fear. Since most of these assets are rated similar, the banks can sell the best ones, get a high price and then use that book value going forward on the other assets they are holding. And they know sooner or later those asset values will out so in the meantime they undertake risky plays to try and make up that asset value.
Pingback: Agents Looting FDIC while cranking the Vol. « Rortybomb
Great post, at least for this populist but financially illiterate reader. One question: won’t it be the case that in all these scenarios, the hedge funds will essentially bid their profits down to 0, so that the main transfer will be straight from the FDIC to the banks? Or is that not exactly the case, because the hedge funds don’t actually know what these things are worth, and thus their bids won’t necessarily converge on the correct price? In which case, perhaps the Treasury is betting — as per your note (1) — that this is still a panic, and that the buyers will significantly underbid, and therefore the Treasury’s half of the profit will actually counterbalance the FDIC’s losses (which, if I read your chart correctly, are minimized anyway when the buyers significantly underbid).
[Sorry for any misunderstandings here, by the way; I’m entirely new to all of this…]
If you assume away the problem of risk aversion, then of course you end up with your conclusion. The fundamental premise of the plan is that investors are massively risk averse: they require significantly greater rewards for a given level of risk than they did nine months ago. As a result, they’re not going to overbid, even with the leverage. Now, maybe this is wrong. But there’s certainly some reason that investors are only asking for a 2.7% yield to lend moeny to the government for 10 years at a time when the Fed is printing money. That does not suggest that investors are, in any sense, risk-neutral.
Also, according to the logic of this post, most of the loans in America are massively overpriced, since banks fund most of their lending with non-recourse loans (that is, FDIC-insured deposits). Do you really believe that banks are out there just routinely overpaying, indifferent to the possibility that they will lose all their capital?
Isn’t this insanely risky? Doesn’t this plan put the entire FDIC at risk of not being able to pay out deposit insurance?
This is essentially saying, “hey, we’ve got a depression like problem. a bank run of a different kind. let’s use that handy institution that solved the bank run problem again.”
In the meantime pillaging that institutions of its capital reserves, creating legitimate consumer fear that their retail deposits are not safe, and then perhaps causing a true depression-style bank run.
I understand that the FDIC would have to have massive losses for this to happen.
But still. I think this is absolutely crazy.
You lost me at “red” line and “green” line. Why do people insist on using colors in graphs that are problematic to a sizable minority of human beings.
Sorry for the interruption. Please continue with your discussion.
Oh wow, visitors. First off, I’m so sorry Mike B. They usually don’t let me away from the number crunching machines, so I had no idea these graphs could be problematic. Sorry!
Ed/ginandtacos, my suspicion is that the stress tests are coming back far worse than they expected. Everyone talks about how hard it is to value these instruments – it is, but it is not impossible to get a ballpark. These instruments are very sensitive to unemployment, which is ballooning – I imagine any of the them are going to have a terrible 2-4 year time frame. So I can imagine Obama et al understand they need to nationalize and are taking a medium term approach – their asking for additional powers right now feeds into my suspicions.
Alternately, Geithner/Summers/Obama think that our system of banking in place is fundamentally good, and just needs this one mulligan to get it back to health. This worries me greatly, especially given Summers trek from killing derivatives regulations during Clinton to DE Shaw. I’ll talk more about this in a post later.
Thanks for the nice model. I’m actually interested in the matlab code you used to put this together since I’m usually a mathcad user and would like to see how this is done in matlab.
My suggestion to Obama would be to put a Tobin tax on financial transactions (a small percentage on the trillions that are traded each day), do their stress test to see which ones need recapitalization. Give haircuts to creditors of these banks, wipe out the equity holders and recapitalize the banks under a receivership using money from the financial transactions tax.
K. Williams – ah yes! Risk Aversion. Often a problem for financial engineering, remembering what metric space you are in (have you ever used Girsanov’s theorem to go from risk averse to risk neutral metric space early in the morning – it smells like napalm!)
I want to keep this accessible to non-finance people (and I’m writing this late), but I’ll kick it up for a second, because it gets to an important point. I did mention this by saying ” and with the government’s backing, we are “risk neutral”, and can bid the straight average.” Getting bidders to risk neutrality is the point of this plan, agreed. So I’m not sure what you’d disagree with here – I agree that only with a massive bailout would people approach these assets.
1) Option payouts have nothing to do with risk premiums (that is factored in with the stock price/vol, when it comes to Black-Scholes); the two blue lines of the payouts for the bid have nothing to do with what we expect nor how adverse we are.
2) I take Brad Delong very seriously, so when he says in the FAQ “That way they would be willing to buy assets at 4/3 of their expected hold to maturity value *if* they were risk neutral, less since they are risk averse. This way I’d guess that they are willing to pay double their estimate of the expected hold to maturity value, because of the value of the Geithner put” I agree – though I look at scenarios where they bid the expectation, which is risk averse in the sense that it is not the expected value WITH the option.
3) In addition to the fact that I hate to lose $100 more than I like to find $100, I am risk adverse in so much as there is Noise Trader Risk (Delong!) and Limits to Arbitrage – there is a good chance that the market can be irrational longer than I can remain solvent. Making a public-private zombie avoids this – since the government prints money, and has a long time frame, piggy-backing my risk aversion function onto their takes care of these short-term worries. This is Delong’s point, which I tried to emphasize. Which is to say, the Put Value both gives me a payout and adjusts my tolerance for risk. Absolutely true.
4) So in that sense, risk neutrality is a function of joining forces with the government zombie. But that said – is it crazy to consider some investors are risk neutral? Warren Buffet? Long Term Value Investors? Private Equity? Sovereign Funds? Hedge Funds? There are a lot of people in the investment world, and if there was value in investing in these given that I can take some hits, I imagine someone would have gathered it. But that is not the case….
As for FDIC, exercising this put option requires a major transaction fee – namely you lose your job, and your company. The hedge funders won’t have to worry about that – their put option is as difficult to exercise as typing “max(X,Y)” into excel. See the next point, and the interfluidty it links to, for more.
Thanks for an excellent comment!
JD – You are very correct. Good intuitions! If they bid down to zero (I had 2cents in the chart, didn’t fix that well) then FDIC gets the most hosed. Picture that chart with the red line running to the right.
If they seriously underbid, there is a very real chance the bank won’t take the bid. This is great if all this is a quarterback sneak to nationalize by Obama, a disaster otherwise.
But you are correct – Hedge funds bidding will bid towards zero profits. They might not make it because (a) they are risk adverse, see the clusterfuck thing I just wrote above, and want to have some cushion in case they overguessed and (b) there are only a few bids, and hedge funds think that they oligarchy bid on it to max their own profits collectively.
They may, and this is important, also decide that it is more profitable to use the asset as a hedge to drive up the value of their Geithner Put by running up the vol; check the next post for more. In this case, they really just want an expected value of zero from this asset, while running up the volatility of the portfolio they select to bid on. This will be interesting.
Well done, but…
The 85% leverage feature got lot of attention based on the early leaks to the Times, and Krugman (to pick an influential name seemingly at random) has spoken of nothing else.
However, the program has a Legacy Securities wing and a Legacy Assets wing (as well as TALF).
“Legacy Securities” is targeted to be about half the program and would cover asset backed securities. The leverage would be 33%, and on an exceptional basis they would go to 50%.
One might presume that if the asset-backeds include junior and intermediate tranches of CDOs that the security could go to zero even if the underlying real estate asset did not, so even with the reduced leverage the non-recourse feature has value. However, its value is much less than at 85%.
On the Legacy Assets program the *maximum* leverage is 85% (Ok, 6-1), based on FDIC approval and review of the underlying assets, which will be *non-securitized* whole loans based on real estate. Unless the real estate project is literally on a toxic waste site or both the house and the underlying land have somehow burned down, those real estate loans won’t go to zero. Let’s hope the FDIC can pick up on those toxic loans on toxic sites.
So sure, a non-recourse loan embeds a put option. The value of that put depends on the allowable leverage and the characteristics of the underlying assets, about which we know little but which are subject to FDIC review. In other words,the same good people who wold oversee a bank nationalization and good bank/bad bank revival are overseeing this.
I would also flag a different issue – private investors cannot create inflation to effectively repudiate their debts; the government can. The Treasury/Fed/FDIC are putting themselves in a position where a little inflation would be a good thing. And sometimes it is! But now the referee of inflation is also betting on the game.
Check out this:
How to divest of toxic assets (China Daily, March 9, 2009)
http://zoudingxie.blog.sohu.com/111139854.html
Miike, Great post!! You’re much better at this than I am. I tried to grind through a similar analysis with Excel but hit the wall….
What I was trying to figure out was something slightly different. It’s transparently obvious that the FDIC can get hosed. In fact, it seemed so bad that I had to assume that the leverage offered by the FDIC would need to vary based on the quality of the pool (unless Treasury is completely in the bag for Wall Street, which is possible). There is some language in the news about the leverage being ‘up to 6:1’ so for higher quality pools you’d get 6x, for lower quality only 1x, or something.
I tried to model 4 pools with different distribution characteristics (really bad, bad, OK, and good) to see what leverage would be required for the hedge fund to get a specific IRR.
From this, we could see how aggressive the FDIC was. They’re the one with the dial that they can turn up or down and deliver alpha to participants. In fact, as this program proceeds, by looking at the leverage offered, we’d be able to determine exactly what they’re offering participants, how much of a wealth transfer this is, and who Treasury is really looking out for.
Pingback: The Glittering Eye » Blog Archive » Looting the FDIC
you’re missing a critical element — the treasury matches investor equity 1 for 1
so 6:1 leverage from the FDIC
.5 of the equity comes from private sources
.5 comes from Treasury TARP funds
this changes the payout ratios significantly if you consider Treasury and the FDIC to be the same entity (ie USG)
in essence the government is selling a free put option to itself — or in other words, it is monetizing the full faith and credit of the USG, letting it all ride on bets on junky assets
What about the cash flows/interest payments?
The securities are unlike options in that they generate cash flow for investors and debt service. If you are buying the bonds at .35 and the original yield on the portfolio was 7%, that is about 20%. I don’t see where your model accounts for this.
Pingback: That’s the signpost up ahead - your next stop, the Financial Zone! « The Long Goodbye
“What about the cash flows/interest payments?”
Good point. Another way for the Treasury to protect itself is to restrict dividends from the PPIF (ie., the investment pool.)
If the equity holders must direct 50% of cash flow to paying down the FDIC debt, for example, the leverage drops pretty quickly from wherever it started.
Or, if some fraction of cash flow must stay in the pool invested in new assets, leverage still declines over time, just more slowly.
Well. That would be subject to negotiation with investors, and it only mitigates, not eliminates, the non-recourse issue.
It seems to me if you include the 20% annual return into the investment calculation, it makes it much less likely that the investment will lose money for either the equity or the debt holders, invalidating your analysis.
Thanks all for the great comments! I wrote an update above to talk about leveraging, and what if the assets turn out to be much better than we expect.
I know non-recourse loan as put option is self-evident to a lot of people, but I wanted this to be accessible to some degree for those who maybe freak out at compounding interest. I tried.
Sammy I made these investors hyper-rational and brilliant, so they know the distribution the assets are drawn from in advance; the bid in terms of $100 already includes the discounted annual return. So whatever the expected return is, it is normalized to $100.
There are a lot of ways to tilt this in favor of the government recoup – restricting dividends like Maguire says. We’ll see if it happens!
Steve I brought it up in comment 1. I had already made the charts into jpgs, and didn’t want to redo it. I do think that “from the government” isn’t always the best frame of reference, something I argue in my latest post.
Matlab code posted!
https://rortybomb.wordpress.com/MatlabCode/
I have a $1 Billion house tape you can analyze to get some feeling for real values.
Mike,
I apologize if I am being dense, and I may well be missing something in what you are saying. But if I am right, it could explain why the FDIC fares so poorly in your model.
I sometimes buy distressed bonds, and the yield is an important part of the return.
Suppose I buy a face value $1000 10% bond for $500 (at that price it yields 20%). One year later, for whatever reason, the bond is worth $400, so I have lost $100. This is where your model stops and the FDIC begins taking losses.
However, we bought a bond not an option, so we have also recieved an interest payment of $100, so we have not lost but have broken even, and the FDIC is still whole and covered. Hopefully we will do better next year.
The point is: the current return on the portfolio can offset principle losses making it less likely for the investor to be wiped out and the lender to take losses.
Duncan – I would love to analyze some housing data, how public access is this?
Sammy – no problem, I’m getting confused too trying to explain so much with this! I’m totally doing a sneak there – I’m just taking the expected discount of everything and normalizing it to 100. So add up all the interest payments, prepayment options, credit risk, etc. etc., and then normalize it to 100. It’s a cheat, but as I said I don’t have any interest in actually trying to calibrate this distribution to the distributions of outcomes – I more want to see the dynamics of the option’s value.
“the real issue with determining this in this tail median/mean issue”
Yup. That’s the biggie. I don’t think your expected distribution is right, and that as a result hedgies won’t overpay as much. But your basic premise works: FDIC gets hosed. What you should have calculated is the bid price for an HF return of 10% — that’s likely what they’ll want, maybe a little higher for doing the valuation work. If I guesstimate that the bid comes in as high as $36.64 for 10% (I’ve got no idea, but let’s say). That’s a 20% overpayment. Ouch. But the sting is that if bank wants $80, true market price is $30, you’re still only up to $36. The bank sniffs and walks away and then whaddya have? (Unless the “free money” angle leads to even much more massive overpayment …) 🙂
You know though, a possibility is that this is sort of drecky, but not horrible assets, and true value lies between 40 and 60. Then I think you go much more bell curve and this HF advantage washes out. But, as you rightly note, it depends on where the truth lies … hmm. I tend to agree it lies on the 40 or below end, despite the bank protests.
To Sammy,
What you say about interest makes sense, but do we know how interest payments will be distributed? I would imagine this would make a huge difference on the value of the security, possibly making it much worse for the taxpayer if the Gov only sees cash once principal is returned. I haven’t looked through the actual plan’s wording, so if anyone has and could shed some light it would be much appreciated.
Pingback: Why are we not yelling about Geithner’s toxic plan? | AccMan
So, I’m a big non-believer in statistics and monte carlo simulations. Assuming normal distys are what got us into this mess in the first place. Clearly many of these outcomes are very dependent. So, I think your “sneak” assumption is where this all goes wrong. It’s a huge assumption and changing it would change the outcome of your model.
I refer you to the following:
http://www.wilmott.com/blogs/eman/index.cfm/2009/1/8/The-Financial-Modelers-Manifesto
Thomas,
The thing of most interest to a general readership, the idea of a giving away a non-recourse loan as giving away a put option, and the graphical representation of the private sector gaining the upside while the public gains the downside, doesn’t depend on statistics. That distribution is drawn on the payout line just as a reference – the final payouts for any asset price just depends on the strike price.
As for the sneak, we’ll see if the subprime portfolio turns out to look like a highly rated valuable corporate bond portfolio.
I’m very familiar with Wilmott’s manifesto and use it to influence my own thinking about my personal code of ethics as a financial engineer. But I’m not sure what you are accusing me of – both here and when I make predictions I’ve made my assumptions clear, stated my methodology, explained my limitations in knowledge, posted my code even, and encouraged the exercise to think of the order of magnitude of the value of the put option rather than a specific value with huge precision. Wilmott clearly is of the opinion that we shouldn’t throw out our math and computers and rely on the Sun and our guts – he wants us to be humble in our approach. I think I did that here.
So as a non-believer in statistics, and I assume modeling, I put the question back to you – how would you approach this problem, estimating how much could the FDIC, and thus the taxpayers, lose? Is it not an interesting question?
Actually the more I thought about it, I realized that my analysis of why your analyis is wrong was flawed. Your analysis is flawed because of the width of the distribution of returns. To say that the returns will be between 0 and 100 and “no one knows” is absurd. The real distribution will be much tighter for most of these securities and knowledgeable investors will actually have a very good idea where they come out. Sure some of the exotic tranches may have flip 0 v. 100 values, but even then, sophisticated investors will be able to get a good handle on it.
So regardless of using normal distros or any distro with such an extreme assumption will lead to the sort of soft analysis that will lead other people who don’t understand finance to call this a “fantastic analysis”. But this is typical when using the “ignorance of the crowds” to assess complicated things.
Also you assume wrong that I’m a non-believer in modeling. Statistics are not necessary for modeling. I know that runs counter the last 10 years of finance, but frankly, that’s why we got into trouble in the first place. Quantum physics and statistics do not a good financial analysis make.
As for this particular question, I think it’s dangerous to model without a deeper understanding of the details of the portfolios and the securities. This is a good example of something being true “in theory” which is not true “in reality”. Economists fall prey to this fallacy all the time. Probably why they call it the dismal science.
How would I approach this? I would get knee-deep in the details and wade through the security documentation to find the underlying securities and then I’d probably go about 4 levels deeper than current modeling by tracing the relationships through all layers of securitization back to the actual underlying assets (mortgages, car loans, etc…). In some cases it’s been reported that the pools are composed of highly related securities (mortgage pools consisting of a few select and highly toxic counties in Miami County, por ejemplo). In those situations, I’d probably make zip code models that incorporate area-specific unemployment numbers, local significant employer lay off tracking, governmental employment rates, analysis of the specific mortgage provider, and “jingle mail” utilization rates. In other situations for other types of security pools, I would go just as deep to develop an understanding their potential payout profiles. (Probably an appropriate level to use statistics.) Then I’d roll it all up back through the layers of securitization focusing on accurate modeling of the payout and claims structures.
But this is probably really hard to summarize so the public can understand what it all means, they like their MTV analysis. Which is fine, because it allows those willing to do the work the opportunity to make money.
Sorry I’m so bitchy about this, but I get steamed when I see the public getting all excited about a simple analysis of a complex issue. The idea that complex things made simple is a fallacy that got us where we are today and is lazy. I guess I’d say I’m peeved when people don’t respect necessary complexity. I know you hide behiind “well, I’m just trying to simplify a hard issue for the people”, but you’re really attacking Geithner’s plan, which, hey, I don’t like Geithner anymore than the next guy, but this attack is actually poorly reasoned. If the distribution of the payouts is actually security specific and one assumes that the investors who put their money in don’t want to lose it and will actually refrain from “prisoners dilemma” bidding, then it could actually be a great mechanism for price discovery. Now, it’s certainly possible that the investors go in and still bid below the banks marks, in which case this is all moot because there’s a much bigger issue at hand (i.e. what do we do when all major banks are fundamentally undercapitalized? Although, at that point, we’ll all start talking about just lowering the required capitalization ratios (watch)).
For another data-point about how bad modeling creates bad results, read the excellent article put out by Wired Magazine: “Recipe for Disaster: The Formula that Ate Wall Street”.
http://www.wired.com/techbiz/it/magazine/17-03/wp_quant?currentPage=all
Thomas – No problem, I appreciate the thoughtful comment.
– Even if you get “knee-deep”, you need to project default likelihood. You say you will do this by “make zip code models that incorporate area-specific…[lots of data] rates.” Fine. Isn’t that what the ratings agencies and the investment bankers did on their mortgage portfolios before slicing them up into CDOs? Make some historical correlation matrix on the zip codes? Heh. I’m sure that’s what the hedge funds will have to do, but it didn’t work out that great the first time around.
– See I don’t like the condescension about “MTV analysis” and how we just need to trust those who are doing the hard work in the trenches. I didn’t like it when the CDS OTC market started getting big, and everyone and myself thought this was concentrating huge amounts of risk, and we were told “Ummm, you aren’t actually looking at the books at AIG – they know what they are doing just fine. How would regulators know their books better than them?”
The public and the government needs to bring a more critical eye to what goes on in their financial sector. The attitude “the people doing these contracts know it better than anyone – I mean, they are knee-deep in it!” has been a complete failure.
– “and one assumes that the investors who put their money in don’t want to lose it” = if that’s the case, they’ll have no problem not taking a non-recourse loan for their position. I doubt that, and will not give them the benefit of the doubt when the relationship between variance and a put option is so elementary. They have every incentive to overbid, and I do believe in incentives.
– “we’ll all start talking about just lowering the required capitalization ratios.” Yes! I am worried about that too. Do you really think the distribution will be “tight” and the variance low on a portfolio of distressed consumer loans with U3 approaching double-digits? We’ll find out!
– But you still didn’t answer my question – to get the data for these securities one needs to be able to raise millions/billions of dollars according to the plan. I’m not going to do that, and even if I could I don’t have the time to analysis it. Given that if the plan costs FDIC only ~$30b, I am ok with it, but if it closer to ~$200b I am calling my congressman to complain, what should I as a citizen who wants to be informed do? “Trust the experts who are knee-deep” is not a viable solution for me, especially now, and especially with the investor’s rational incentive to game the loans and increase their volatility.
No. PPIP is not designed to encourage discovery. It is designed to reinflate a collapsing asset bubble using public monies. Meaning it’s an attempt to *postpone* the final collapse of the bubble long enough to transfer assets to the government, so that the government — rather than the banks — can absorb more of the losses. Any “discovery” that does happen to take place will be thoroughly bogus.
I’m happy to accept your first and second premises and leave the third one unresolved, and still argue that the plan as presented is an explicit invitation to overbid.
If you don’t like hypothetical lotteries I can provide a strictly agent-based explanation of why that is. No statistics required. I’m guessing it’ll take about 800 words though.
Radish,
I’d be curious as to an agent-based explanation. I played around with some game theoretical stuff on what state-spaces will the banks accept or decline the auction bid, but all of them assume that the bid from these PPIP processes will be greater than a regular bid.
Feel free to post it here, or elsewhere and link to it (probably better). I’ll throw in my two cents….
Hmm. I keep meaning to set up a public blog but haven’t gotten round to it yet, so I either clutter up people’s comment threads or post at community places. Lemme see what I can do in the way of a community-suitable post about why it isn’t plausible that this about price discovery.
Assuming of course that the folks at Treasury have actually thought through what they’re doing, instead of just acting out of blind panic 😉
Here ya go. Looks like I got a little carried away, though…
Very nice presentation … suggestions from first impressions for illustration clarity:
Some heuristic explanation of Monte Carlo simulation would be helpful in making the leap from the option payoff graph to the matrix of pricing scenarios. For example, include a little more description of the outcome from base case pricing at expectation. The description as it stands is a little terse and short cut: “However, looking at the second line, since our downside is covered, we transfer out $3.66 for putting up $4.38, a nice 83% profit transferred straight out of the FDIC, which loses exactly that $3.66.“ This takes some re-reading even for those of us with a bit of knowledge of options but who are out of practice. But the logic of your bid progression process is excellent.
Your model assumes that the market consists of a large number of assets – the expected results must require tens of thousands or more bids to get any confidence. But from Geithner’s plan, it is hard to estimate how many pools there will be but difficult to imagine that there will even be thousands.
monte-carlo simulations over a universe of 100 or 200 pools would give nobody any real confidence.
Pingback: The Geithner Put - The Public Private Investors Partnership Fund — OptionPundit
Pingback: Save the GOP » Blog Archive » Bad Idea Overload
We need to get back on the gold standard and stop this derivative foolishness.
The Centrals Banks are Wall Street and use these logarithmic projections and toxic assets to do exactly what the observant contributor mentioned earlier in this column, Money laundering with Wall Street and the Central banks( soon to be singular as in Central Bank) the principal benefactors of this
otherwise illegal and immoral bit of prestidigitation. Watch me pull a rabbit and a toxic asset out of my hat!! They should be arrested on charges of treason and all hung from the highest mighty oak.