Sunday, March 29, 2009

Overpaying Under the Geithner Plan: A Tough Nut to Crack

Warning: This is LOOOONG and a bit GEEKY.

It's been almost a week since Treasury Secretary Geithner took the wraps off his plan to rescue the U.S. banking industry from an onerous backlog of toxic assets. Geithner sketched out a scheme (PPIP, for short) wherein public and private investors would buy assets jointly, with the FDIC kicking in a large non-recourse loan. That fat FDIC loan sweetens the deal for private investors (such as hedge funds) because (1) the FDIC loan is very low-interest (2) The FDIC eats all the deep losses (the public and private investors take the “first loss”).

Immediately the financial engineering-minded types started to sort through the problem of how much overpaying would result. It’s been a wacky week in blogland. Estimates are coming in all over the place, based on differing approaches, differing assumptions. I like the work done by Rortybomb, and Nemo (after recovering from following Krugman off a cliff with a badly oversimplified model) has produced some good stuff too (his comment section is worth the trip -- some really zingy and smart chatter on the sidelines).

Anyway Rortybomb’s latest estimate: there will be overpaying in the low teens, say about 12 percent. This may be correct, but I’m reaching the “throw up my hands” point. That’s what today’s entry is all about: broadly, why it’s so damn hard to get a grip on the magnitude of the overpaying. Let's go, point by point.

1. These assets are hard to value in the first place, even in a good economy.

The so-called legacy securities are complex, currently illiquid and basically unique. (I read once there are 100,000 different types out there clogging up bank balance sheets. Incidentally, for this reason the reverse auction idea was abandoned -- that approach works great to sell commodities like road salt, but not well at all for collections of complicated, unique products.)

2. In a bad economy, with high market volatility and a very uncertain future for growth and unemployment, placing a value on the assets becomes even trickier.

3. Once you value the security, you have to make adjustments for the “FDIC effect.” In other words, you're buying it through the Geithner special program where the FDIC shields you from deep losses and provides low-cost money. This effectively enables you to bid higher. But how much? Part I: calculating the value of the “embedded put.”

There’s been a lot of talk about the “put” in this plan. In this context, “put” refers to a kind of option commonly used in financial markets. It’s basically a bet that a price will fall. It gives the owner the right to sell a security for a certain amount, within a certain time frame. Confusing to novices, I know, so let’s look at a real world example using shares (“puts” are common on the stock markets).

Let's say I think Microsoft stock is going to plunge, so I buy a $3 put that expires November 30. It allows me to sell 100 shares for $30. (Say they currently trade at $35.) Summer comes, the new Xbox Millennium Plus is a stinkeroo, and the shares sink to $25. I exercise my put (I buy 100 shares at $25 each, then resell them to the writer of my option for $30). My profit: $5 a share (the difference between $30 and $25). I pocket a cool $500.

Now how does the Geithner plan contain a put? The plan works like this, according to one example provided by the Treasury Department: if the bid on an asset is $84 (out of a face value of $100), the private investor chips in $6, the Treasury throws in $6, and the remaining $72 takes the form of an FDIC non-recourse loan. The private investor won't lose more than what he contributes: $6. So he is protected from the worst ravages of downside risk.

It's like having a put option on $72 ($72 not $78 because he shares losses on the way down equally with his public partner, so his six dollars don’t evaporate completely until the price hits $72). More accurately, it’s a put embedded in the security (and so is a little less valuable -- in a perfect world, you could strip out the put and trade the pieces separately. For instance you could sell the riskier, high-leverage put to a hedge fund, and the asset itself to a more conservative investor, say a pension fund).

So the point is: you also have to figure out how much this put is worth for the particular asset you’re bidding on.

4. Part II: Calculating the value of the low-interest loan.

The FDIC supplies a low-interest loan. That loan, especially over a multi-year asset, could be worth quite a lot.

Let’s say the private investor, a hedge fund, demands a 12 percent return on its investments. Also right now, in these tight credit markets, it must pay 15 percent to borrow large quantities of money (here lies the liquidity problem). So how much would it pay for a five-year asset predicted to churn out yearly revenue of $27 million, then return a lump-sum principal of $100 million at the end?

This is an easy one: $100 million (for simplicity, I disregard inflation). Every year it pulls in $27 million ($15 million a year covers its borrowing costs, leaving $12 million for profit). At the end of five years, it gets back the $100 million.

Now what happens if you substitute a 2 percent cost of borrowing (let’s pretend for the sake of simplicity that the FDIC loan is about 1 percent and the hedge fund’s own costs bump the average up to 2 percent)? How much would the hedge fund pay for the asset now? Almost $145 million, or 45 percent more!!!!

(For those who want to double-check my math: it doesn't simplify all that easily and I initially got it wrong -- whoops! -- so I'll refer you to the comment section, where a poster straightens it out for me. If anyone out there wants me to revisit this calculation in a later blog entry, just drop in a comment and I will. Else, my hunch is 99 percent of you don't care that much, though the underlying point is quite critical: being able to obtain money much more cheaply than at market rates can really boost the value of a long-term asset.)

Ah, but hold on: I cheated a bit. Yup, I assumed this current distressed credit market will be exactly the same in five years, with that burdensome 15 percent borrowing cost unchanged. That’s not likely. Factoring in lower borrowing rates will push down the level of overpayment.

There's also arguably a problem with my initial assumption of a 15 percent borrowing cost. On one hand, we hear that the costs of loans have soared, but then again, private investors are reportedly saying they have enough money and are champing at the bit to have at these distressed assets ... also the broader interest rate environment in the U.S. is close to zero right now, so 15 percent may not be correct at all.

For kicks, let’s change it to a 4 percent borrowing cost instead of 15 and tweak the numbers slightly to keep the math simple. Let’s assume the asset kicks out $16 million a year over five years, and again returns $100 million in principal. Got it?

So now the hedge fund would pay $100 million ($4 million a year will cover borrowing and $12 million profit). Now, lower your cost of borrowing to 2 percent because of the FDIC subsidy, and you end up overpaying ... less than 7 percent. Bad, but not nearly as bad as 45 percent. Quite a big swing in fact.

Now, what’s the reality? I honestly don’t have a good feel for this, but it’s important in figuring out the overpayment.

Anyway the bottom line remains: your bid is affected to a certain degree by the advantageous cost of borrowing through the FDIC non-recourse loan.

5. Then other questions are swirling about that will affect your bid, some practical:

Valuing complex securities can be time-consuming and expensive. You’ll have to build into your bid a percentage point or two to cover the costs of doing a lot of research but then losing out to other bidders (or having the bank turn down your bid). This will nudge your offering price a little lower.

Also, are you allowed to resell the asset into the secondary market? How would that work? Here’s the danger: You buy a $100 million five-year asset for let’s say $130 million because of the value of the FDIC subsidy and the put, then need to resell six months later. But does the buyer automatically get your FDIC subsidy and put transferred to him? It’s a problem either way. If he doesn’t, he’ll want to bid your asset closer to $100 million and you’ll refuse to sell. So in essence you’ll be married to an illiquid five-year asset (which will tend to push your initial bid lower). If he does, then the subsidy is built into the asset. That kind of approach isn’t exactly market-friendly -- the asset doesn’t get a temporary subsidy; it gets a lifelong subsidy. What’s more such a provision can distort the larger markets; if the subsidy effect is too large, it will serve to make it harder to discover true prices in the market at large.

6. And some intangibles may affect your bid:

Such as: Is this plan going to turn around and change in midstream (this is the government after all)? Are you going to be demonized for seeking a big profit? For an investor will it be a lose-lose proposition in the court of public opinion? (If you profit big, the public will hate you for doing so well. If you overpay big and the taxpayer loses big, the public will hate you for bailing out the banks.)

Okay, almost done! One note: I’m not saying the bidders will actually break down the elements of their offer this way. I’ve simply done it like this to show how many parts can be teased out of one simple bid. And they all affect price, to greater and lesser degrees, and so would be factored in.

After sketching out all these complexities, I have to admit the kicker is that all this analysis may be for nought. What could be the biggest influence on price is gaming the system, if loopholes exist. If I’m a bank that owns a private equity outfit, and I can arrange for a full-price bid, say $80, on crappy assets worth $30, why not? FDIC eats most of the losses. If the Geithner plan hasn’t taken steps to ensure arm’s length transactions, members of the Obama administration are very stupid and possibly criminal. If that worst case comes to pass (i.e., massive gaming of the system), there will be blood on the streets when the public finds out how they got fleeced by their own government.

Update: Thanks to the anonymous commenter below, I fixed my math (the overpaying under part 4. turns out to be 145 percent, not 138 percent ... I also rejiggered the other example in that section to be correct.)


  1. I think you assumed something wrong in your calculation or the description of the asset in 4. Part II, cause you didn't add the principal payment of 100 million at the end of the five-year asset.

    If the total payment is 135 million instead of 235, then the price would be around 70 million, not 100 million. If this is the case (235 million of total payments), then a borrowing rate of 15% for a YTM of 27% would correspond to a price of 100 million. If you decrease the borrowing rate to 2% the price of the asset would be 145 million, not 138 million.

    I just thought you needed to clarify this in order to follow your post.

    I would appreciate if you could comment on this issue

  2. Hmmm. I was mainly trying to illustrate the point that assumptions about borrowing costs can lead to a huge swing in estimates of overpayment premiums. But there is a problem with my simplification, you’re right. Let me review all this (my writing was rather sloppy in the first place, so I’ll try to fix that here). See if this makes sense (if it does, I’ll fix this entry and rewrite it somewhat to make the points clearer):

    First, I never set this up well in the first place, so let me do that now. The question: what would you pay for an asset that kicks out an income stream of $27 million a year for five years, then at the end of that period returns a principal of $100 million (this then gives it mortgage-y characteristics)? Let’s assume the FDIC subsidy scenario: you need to make 12 percent profit yearly but your borrowing costs are only 2 percent.

    The example I use above says $138 million and here’s why: It takes 14 percent a year, or $19.32 million, to cover profit + cost of money. That totals $96.6 million over five years, against $135 million that the income stream kicks out. The difference is roughly $38 million. That’s what you need to make up for the “loss” on the underlying principal -- it’s only worth $100 million after all, but you paid $138 million. So paying $138 million leaves you whole at the end of the five years.

    But this isn’t actually the full picture (this may be what Anonymous is referring to): Each year you’re getting $27 million but your required profit + cost of money is only $19.32 million. Now, assume that you invest that extra money at 12 percent (that’s the typical yield of your investments), you’ll wind up with more money in pocket after five years and thus be willing to pay more for the asset up front. I scratched out quick calculations that got me to about $143 million or so, in which case $145 million seems about right. (NOTE: I wasn’t using any kind of mortgage calculator here, so it probably makes sense that I’ve gone astray.)

    Anonymous, please let me know if this is basically what you’re looking at. If not, please show your work (try to keep it simple) for reaching $145 million and, as long as it’s right, I’ll update the blog entry with that figure. Thanks.

  3. Thank you for replying to my post. I like your blog and the fact that is written for people to understand current issues with no financial background.

    The 145 million figure was obtained using simple present value of future cash flow calculations, which assume that investment of cash flows prior to maturity (the payments of 27 million yearly for each year,except for the last) are actually reinvested at the required rate of return of 14%. Assuming nothing else changes until maturity of the assets, the 27 million yearly income can be reinvested at 14% because prices of the assets under the program would imply this rate during the life of the asset under equilibrium conditions.

    In order to calculate the increase in prices due to the lower borrowing costs we should calculate the price under the assumption that the non-recourse feature in the loan made by the FDIC is not provided. The fact that is non-recourse gives investors a free put option, as explained in your post. The total amount of "overpayment" will be related to both the put and the subsidized interest on the FDIC loan.

    Here's a brief of my calculations, based on the fact that the price is 144.6 million (around 145 million). If this is the case, then 20.2 million would cover cost of borrowing + profit. Therefore, 6.7 million would be left in order to reinvest. The last can't be reinvested, so its value at maturity is 6.7. Nevertheless, 6.7 reinvested at 14% for 1 years (which would correspond to the next to last payment of 27 million) would have a future value of 7.7 at maturity. Likewise, the earlier interest income payments would have future values at maturity of 8.78,10,and 11.4. Adding these values results in 44.6. When you add the 100 principal payment, the total price would increase to 144.6. This increase in prices fueled by lower borrowing costs are currently one of the reasons set forth in the media to blame Greenspan for the asset bubble.

    So, you're response is basically correct, although the reivestment of interest income should be at a rate of 14% in order to match the risk of the original investment.

    I hope I made myself clear, and would like to hear from you again.

  4. Well done! I back-checked your math and it looks impeccable. Indeed, in simplifying I overlooked this piece of the calculation. I humbly concede the high ground and will take this opportunity to do a little surgery on this blog post. Good point too about the lower borrowing costs helping fuel the housing bubble -- that’s very true, and this example we’ve just banged through (though maybe a bit hard to follow for a non-mathematical sort) illustrates that well. Thanks for stopping by and correcting this.

  5. I appreciate the time you spent in order to double check my math in order to refine the details of your post (specially for the few of us that tried to figure out the math in the first place). I agree that most people won't care much about the specific numbers, which are not crucial for the point you're trying to make. Although oversimplified, your comment illustrates very well the issues at stake here, which will almost certainly are going to have immense costs to taxpayers in the future.

    Although the plan is politically feasible, this plan will be touted when the next crisis hits, as it reinforces the perverse incentives given to the banking industry.

    To provide an example much more related to my own experience in terms of its effects, Mexico's response to the 1994 banking crisis had striking similarities (in substance) with the current Geithner plan. Basically, bailing out bankers at the expense of taxpayers, worsened by corruption and lack of market-type solutions for price discovery (although the implicit subsidy provided by the Geithner plan can't be considered market-based either). As a result, we are still paying for the losses incurred, which has proven to be a huge obstacle to Mexico's growth and development.

    I just HOPE that the plan proves to be successful, for both USA and Mexico citizens alike.

    Thank you.